Date of first round report: 1 September 2016 Date of second round report: 30 January 2017



Download 8.29 Mb.
Page41/53
Date conversion16.05.2018
Size8.29 Mb.
1   ...   37   38   39   40   41   42   43   44   ...   53

Second round evaluation

  1. Second round clinical evaluator introductory comments


From review of the first round CER and the response including errata documentation, it is clear that there are significant process-related problems highlighted by this submission. These may be attributable, in part, to the growing complexity of (particularly oncology) clinical trial design and data, and the changing regulatory environment around ‘early approvals’.

Whilst a small number of previous submissions have been recently approved on the basis of early data, it is important to note that the approval of a medicine for a last-line usage in a rare indication on the basis of surrogate markers or uncontrolled data, with caveats in the PI regarding the limited nature of the evidence (and with conditions of approval that include submission of full CSRs for Phase III confirmatory studies when available) is a very different thing to the approval of a medicine for a first-line usage in a very common indication on the basis of a top-level summary document. Whilst summaries are valuable overviews, the review of low-level data and systemic examination of investigative processes is the critical responsibility of a regulator. The TGA is a distinct entity to other regulators, and whilst their stances and reasoning provide collegiate support, their decisions do not dictate those of the Australian government.

The sponsor have submitted an application for first-line use of a first-in-class new chemical entity on the basis of a ‘Topline Summary’ (that is, not a complete CSR) from a Phase III RCT, and supported by data from a study not adequately designed to support a first-line indication. The evaluation of this submission resulted in extensive questions to the sponsor from the first round clinical evaluator.

At this late stage in the application process, in order to support the efficacy and safety claims and ostensibly in answer to questions, the full CSR from the Phase III RCT (Study 1008) has been submitted. It was available months ago, but due to the nature of the current process and the fact that the sponsor had specifically been told not to submit ‘rolling data’, it was not submitted at that time. The second round evaluation therefore involves the review of a large number of responses, in addition to the review of an entire CSR for the pivotal trial supporting first-line registration.

Due to these unusual circumstances, the second round evaluation will focus entirely on the answers to questions posed by the first round clinical evaluator, and review of the PI/CMI documentation, while the CSR for Study 1008 will be reviewed separately by the Delegate. Where findings from Study 1008 are referred to in answers to questions or cited in PI text, this will be taken at face value and cross referenced for review by the Delegate.

Second round evaluator comments will be highlighted in the same formatting as this grey comment box throughout the second round CER.


    1. First round evaluation errata

      1. Errors of fact


An extensive listing of errors of fact that the sponsor has identified in the first round CER is included with the submission. This listing will be considered by the Delegate in constructing their overview.
    1. Review of responses to clinical and PK/PD questions

      1. Pharmacokinetics (PK) questions/responses

        1. PK question 1


QUESTION 1

Although Study A548103213 examined dose proportionality between 4 single dose levels of palbociclib (75mg, 100 mg, 125 mg or 150 – 200 mg final Phase III capsule), no studies have been provided that examine the BE of these 3 dosage strengths nor has the sponsor applied for a waiver for the required studies. Can the sponsor please comment?
Sponsor Response

The bioequivalence strategy, including the justification supporting the equivalence of the different capsule strengths, was presented in Module 3, Section 3.2.P.2.2 Drug Product, as well as Module 2, Section 2.7.1 Summary of Biopharmaceutics. Palbociclib final Phase III/commercial free base capsules are available in 75mg, 100 mg, and 125mg dose strengths. These three capsule strengths have the same qualitative composition, are quantitatively proportional formulations (ie they derive from a ‘common blend’), and are produced following the same manufacturing processes.

To support the equivalence of the different capsule strength formulations, a statistical analysis was conducted on the palbociclib dose-normalized exposure parameters of AUCinf and Cmax following administration of the 75, 100, and 125 mg capsules under fed conditions to healthy Japanese volunteers in Cohort 1 of Study A5481032 (see Module 2, 1023 SCP, Section 2.7.2.1.3.3) following the recommended methodology described in Section 4.1.6 of the EMEA Guideline on the Investigation of Bioequivalence (BE). The results of this analysis indicated that the 90% CIs for the ratios of each comparison were all within the 80% to 125% BE range (see Module 2, 1023 SCP, Section 2.7.2.2.3.1.7.3). Since it has been established previously that palbociclib exhibits linear PK (see Module 2, 1023 SCP, Sections 2.7.2.3.1.2 and 2.7.2.3.1.3) across the dose range of 25 to 225 mg, the bioequivalence between the three capsule strengths indicates the capsule strength equivalence of the final Phase III/commercial free base capsules. The sponsor’s position is that this analysis from Study A5481032 provides the same information as and obviates the need for a dedicated BE study.

Based on the discussion above, the sponsor believes that the information included in the current submission provides adequate support to the dose strengths equivalence and no additional BE study is needed.

‘Common blend’ refers to ‘a batch of final blend that can be packed in different amounts providing various strengths of the capsule product’.14

The question posed by the PK reviewer relates to the difference between bioequivalence (equivalent rate and extent of absorption after administration of the same molar dose) and dose proportionality, which as stated has been shown in Study A5481032.

Per section 4.1.6 of the EMA guideline on bioequivalence:15



If several strengths of a test product are applied for, it may be sufficient to establish bioequivalence at only one or two strengths, depending on the proportionality in composition between the different strengths and other product related issues described below. The strength(s) to evaluate depends on the linearity in pharmacokinetics of the active substance. In the context of this guideline, pharmacokinetics is considered to be linear if the difference in dose-adjusted mean AUCs is no more than 25% when comparing the studied strength (or strength in the planned bioequivalence study) and the strength(s) for which a waiver is considered.



If bioequivalence has been demonstrated at the strength(s) that are most sensitive to detect a potential difference between products, in vivo bioequivalence studies for the other strength(s) can be waived.

A sufficient number of subjects were included in Study A5481032 per the EMA guideline (13 were studied in these dosage groups, and the guideline specifies not less than 12 should be included) and under the correct conditions – that is, fed conditions per the dosing recommendations.

Dose proportionality has been shown in Study A5481032, as described, and the PK shown to be linear across the relevant dose range, so dedicated BE studies are not required.


Evaluator comment:

The sponsor’s response is accepted.
      1. PK question 2


QUESTION 2

As M22 is the most abundant circulating metabolite (responsible for 14.8% of circulating radioactivity), does the sponsor have information regarding its activity?


Sponsor Response


In humans, the glucuronide conjugate of palbociclib (M22) accounted for 14.8% of plasma total radioactivity AUC, but constituted for only 1.5% of dose in urine (PD-0332991_15Oct13_113146). The small amount of this metabolite recovered in urine limited the feasibility of its isolation and further structural elucidation. In addition, the position of the glucuronide conjugate on the parent molecule could not be identified based on mass spectral analysis, hence the definitive structure of M22 could not be ascertained. As such, the pharmacological activity of M22 for CDK4/6 cannot be predicted by the known structure-activity-relationship. Moreover, it is anticipated that the polar nature of this metabolite will likely impede its ability to distribute into cells, thus attenuating its cellular potency towards the intended target of CDK4/6.

Evaluator comment:


The sponsor’s response is accepted.
      1. PK question 3


QUESTION 3

Can the sponsor please provide the complete clinical trial report for Study A5481013, which examined the effects of hepatic impairment on palbociclib PKs?

Sponsor Response


The sponsor has initiated Study A5481013 to investigate the effect of varying degrees of hepatic impairment on single dose palbociclib pharmacokinetics, in otherwise healthy subjects. Upon completion of the study the sponsor will generate and submit the clinical study report (CSR) for this study to TGA. Pending sufficient results, the sponsor will also submit palbociclib treatment recommendations for patients with moderate and severe hepatic impairment at that time. Based on current enrollment projections, it is expected the CSR for Study A5481013 will be available by December 2017.

Evaluator comment:


The PI reflects the current knowledge around hepatic impairment. The sponsor’s response is accepted.
    1. Pharmacodynamics (PD) questions/responses

      1. PD question 1


QUESTION 1

Given that at the mean and median c following QD dosing with 125 mg palbociclib the upper bounds of the 90%CIs for QTcS range from +8.72 to +9.16 msec and therefore are relatively close to the 10 msec threshold,16 is it possible that co-administered drugs that increase palbociclib exposure even by as little as 20% to 30% will possibly result in major safety concerns?

Sponsor Response


In a Phase I dose-escalation trial in 74 patients with advanced solid malignant tumors (Study A5481001), palbociclib daily doses ranging from 25 mg QD to 225 mg QD were investigated with 2 dosing schedules, Schedule 3/1 (palbociclib taken orally QD for 3 weeks followed by 1 week off treatment) and Schedule 2/1 (palbociclib taken orally QD for 2 weeks followed by 1 week off treatment). The recommended Phase II doses (RP2Ds) were determined to be 125 mg QD on Schedule 3/1 and 200 mg QD on Schedule 2/1. As described in Section 2.7.2.3.1.3 of the Summary of Clinical Pharmacology (SCP) and detailed in Table 1 of this response, the exposure as assessed by plasma AUC0-10 and Cmax increased in a dose-proportional manner over the entire dose range following single dose (Day 1) and multiple doses (Day 8) of palbociclib. Relative to 125 mg with Schedule 3/1, the exposure levels at steady state for 200 and 225 mg with Schedule 2/1 were generally consistent with the 60% and 80% higher daily doses, respectively. The ECG and PK data collected in these two cohorts of patients should provide sufficient coverage for assessing the potential for palbociclib-induced QTc prolongation in high exposure level in special populations (for example, elderly patients with low body weight, patients on concomitant moderate CYP3A inhibitor) under the current clinical dosing regimen of 125 mg with Schedule 3/1. Analysis of the QTcF (Fridericia’s correction) and QTcB (Bazett’s correction) versus RR plots revealed that Fridericia’s was the more appropriate correction method, and QTcS (study-specific correction) was also calculated for evaluations. The ECG results demonstrated no patients had a maximum QTcF or QTcS >500 msec on study. Two patients, one in the 75 mg QD Schedule 3/1 dosing cohort and one in the 125 mg QD Schedule 3/1 dosing cohort, had maximum changes from baseline in QTcF or QTcS values of >60 msec. The categorical summaries of QTcF and QTcS are presented across all dosing cohorts in Table 2 and Table 3, respectively [1, 2]. There is no trend of more patients falling into categories of higher absolute QTc values or higher increase from baseline values with higher dose levels. Similar observations were also found in the summary of mean QTc change from baseline by nominal time for each dosing cohort [3].

In a substudy from Study A5481008 that was conducted as the definitive QT interval prolongation evaluation for the palbociclib program, triplicate ECG data were collected at clock time-matched baselines and at 5 time points (pre-dose and 2, 4, 6, and 8 hours postdose) on Day 14 after palbociclib had reached steady-state concentrations following a therapeutic dosing schedule (125 mg QD on Schedule 3/1 in combination with letrozole). Based on these ECG data and time-matched PK data a PK-pharmacodynamic analysis was performed to characterize the relationship between palbociclib exposure and ECG endpoints (RR and QTc intervals) using linear mixed effects model (PMAR-EQDD-A548b-sNDA- 611). While a slight positive linear relationship was observed between palbociclib concentration and QTcS, at the mean steady-state palbociclib Cmax the mean QTcS increase from baseline was 4.04 msec, with the upper bound of the 1-sided 95% CI at 6.67 msec. As seen in Table 4, based on this model the predicted upper bound of the 1-sided 95% CIs for QTcS at 20% and 30% higher than the mean or median steady-state palbociclib Cmax are both less than 10 msec. Similar results were obtained with QTcF and QTcB. Therefore, based on currently available data, there is no expectation of clinically relevant effects of palbociclib on the QT interval in patient populations with higher palbociclib exposure.

Evaluator comment:


The sponsor’s response is accepted.
    1. Efficacy questions/responses

      1. Efficacy clinical question 1


QUESTION 1

Many patients required a dose reduction or delay. It is not discussed whether the clinical benefit for those on a lower dose is comparable with those who manage to stay on the starting dose. The sponsor is requested to provide an analysis of the efficacy outcomes and a forest plot of PFS and ORR to demonstrate the effect by dose received for Study 1023 (with the lowest dose received to be used for any who have had dose reductions).

Sponsor Response


As requested, the sponsor provides Progression-free survival (PFS) and best overall tumor response for three subsets of patients; those who had at least one palbociclib dose reduction during the Study A5481023, those who had a dose reduction to 100 mg/day, and those who had a dose reduction to 75 mg/day. Comparisons cannot be made to the control arm because <5 double-blind placebo-treated patients had a dose reduction of placebo. Thus, a forest plot cannot be generated.

As of 23 October 2015, 132 patients in the palbociclib plus fulvestrant arm had at least one palbociclib dose reduction. Palbociclib dose was reduced from 125 mg/day to 100 mg/day in 89 patients and to 75 mg/day in 43 patients (Table 1).

The requested PFS analyses were conducted in the 3 subsets of patients identified above and are presented in Table 2. A summary of best overall tumor response is also reported for the 3 subsets of patients in Table 3.

As shown in Table 2, the PFS medians of the 3 subsets of patients are comparable. The same conclusion can be drawn from the Objective Response Rate (ORR) results of the 3 subsets of patients (Table 3). The 95% Confidence Intervals (CIs) of the Kaplan-Meier estimates of median PFS and the 95% CIs of ORR for the 3 subgroups overlap indicating a lack of evidence to conclude that they are different.

In conclusion, the sponsor considers that the clinical benefit of patients treated with palbociclib plus fulvestrant was maintained although palbociclib was administered at a reduced dose.

Evaluator comment:


The sponsor’s response is accepted.
      1. Efficacy clinical question 2


QUESTION 2

Study 1003: Please provide the breakdown of the operations as to whether they were breast versus non-breast surgery for each treatment arm. For those who underwent breast surgery, please state the number and percentage going on to receive adjuvant therapy, by treatment arm.

The first round clinical evaluator comments relevant to this question were:



49.1% of patients had de novo metastatic disease which is a much higher figure than the 5-10% that would be expected with Stage IV disease at presentation in Australia.

the ‘prior surgeries’ rate is 81% in both arms, although it is not clear whether this is breast surgery; it would not be usual practice in Australia to perform breast surgery on a woman presenting with metastatic disease and this rate appears very high for palliative procedures. Similarly rates of radiation are 54.8% which may have been adjuvant or palliative.

Sponsor Response


The breakdown of breast vs. non-breast surgery for each treatment arm from Study A5481003 are summarised below:

Table 33: Breast versus non-breast surgery for each treatment arm from Study A5481003



The number and percentage of patients who received adjuvant therapy post breast cancer-related surgery by treatment arm are summarised below:

Table 34: Number and percentage of patients who received adjuvant therapy post breast cancer-related surgery by treatment arm


Evaluator comment:


As the prior non-breast cancer surgery category includes biopsies for breast cancer, it is not surprising that this rate is around half in each arm. The rate of breast cancer surgery is high compared to what the first round evaluator would have expected, it would seem, but is even between arms. Perhaps this high rate was related to the high rate of de novo metastatic disease in this study population as a whole (almost half).

Radiation therapy appears to have been predominantly adjuvant (though it isn’t entirely clear that adjuvant therapy as it is included in the second table refers to adjuvant radiation therapy only).

The sponsor’s response is accepted.

      1. Efficacy clinical question 3


QUESTION 3

Study 1003: Please provide a breakdown of the numbers of the 17.9 % patients for each arm who received no endocrine therapy following an earlier ER-positive breast cancer diagnosis.

The first round clinical evaluator comments relevant to this question were:



66.7% had either relapsed within 12 months of completion of adjuvant treatment or had de novo disease, and these two groups have been put together for stratification purposes. The latter (49.1%) would be expected to have a better prognosis than those relapsing after treatment, which makes this stratification factor likely to lead to prognostic factor imbalances – indeed this did happen with more patients with de novo disease in the palbociclib and letrozole arm;

The rates of prior antihormonal therapy (Table 22, CSR) indicate that 110/165 patients (67%) in the Phase II study received no prior hormonal therapy that is, 17.9% did not receive endocrine therapy following a diagnosis of ER-positive breast cancer – it is standard practice in Australia to offer endocrine therapy to women with ER-positive breast cancer, and may influence baseline response rates to any endocrine therapy commenced in the metastatic setting;

Sponsor Response


The sponsor would like to clarify the number of patients with no prior endocrine therapy is not 17.9% and cannot be derived as described by the evaluator in the Clinical Evaluation Report (page 44) by ‘subtracting the 49.1% de novo group from 67% of the total population with no hormonal therapy’ as 67% represents a subset of patients distinct from those who received prior systemic therapy as displayed in Table 22 of the Study A5481003 CSR. Table 22 is included below for reference.

In Study 1003, there were 57 patients (67.9%) in the palbociclib plus letrozole arm and 53 (65.4%) patients in the letrozole arm who received no prior endocrine therapy.

The number of patients who received no prior endocrine therapy is derived by subtracting the subset of patients who received ‘Antihormonal’ therapy in Table 22 of the CSR from the total number of patients in each arm (n=27 subtracted from N=84 is 57 patients and n=28 subtracted from N=81 is 53 patients).

Of the 57 patients from the palbociclib plus letrozole arm and 53 patients from the letrozole arm with no prior endocrine therapy, there were 13 patients and 16 patients, respectively, who received other systemic therapy. The numbers are summarised below:

Table 35: Patients with no prior systemic therapy and patients who received other systemic therapy

Evaluator comment:


The sponsor’s clarification is accepted.

There were a total of 110 out of 165 subjects (~67%) in the Phase II ITT population for this study who had not been treated with prior hormonal therapy. They made up 67.9% (57/84) of the active arm and 65.4% (53/81) of the comparator arm.



Table 36: Rate of previous or endocrine systemic therapy

Derived from the sponsor’s response, as tabulated above, the rate of previous systemic therapy was higher in the comparator arm by approximately 6.7%. The imbalance of de novo metastatic disease (a confounder also associated with better outcomes) between the arms was identical to the imbalance in previous systemic therapy between arms; these were likely the same individuals.

In the context of the efficacy outcomes reported, where investigator results were borderline and BIRC results gave a hazard ratio crossing 1, such an imbalance in a key prognostic factor introduces further uncertainty about the validity of the findings of trial 1003.

The findings of trial 1003 are not pivotal to registration, nor involved in the information presented in the PI, and thus its flaws are not further relevant.


      1. Efficacy clinical question 4


QUESTION 4

Study 1003: The data presented in the CSR for investigator and BICR censoring respectively, differ from those data presented for the BICR censoring in Table 26, p69 and Table 27, p70 of the FDA report (publicly available on the website) for Study A5481003. The sponsor is requested to provide an explanation for all differences in the data presented in the dossier versus the tables in the FDA report, including but not limited to, the higher AE rates, clinical progression, withdrawal of consent reported in the FDA report noting that the FDA table was generated in response to an FDA query on 28 Feb 2014.

Sponsor Response


The differences between the sponsor and FDA classifications for some of the reasons for censoring in the PFS analyses are highlighted in Table 1 and Table 2. Twelve patients (highlighted in red) were classified as ‘No on-study disease assessments’ as the censoring reason by the sponsor in the PFS analysis based on investigator assessment (Table 1), but were classified as other reasons by the FDA. Similarly, 10 patients (highlighted in red) were classified as ‘No on-study disease assessments’ by the sponsor in the PFS analysis based on BICR data (Table 2), but were classified as other reasons by the FDA.

A patient’s objective disease progression time can be censored for multiple reasons in the PFS analysis depending on what happens to the patient in the study. Sometimes, there can be multiple reasons for censoring. In the programming algorithm for censoring reasons, the sponsor assigns ‘No on-study disease assessments’ with a higher priority if there are multiple reasons for censoring. Patients with ‘No on-study disease assessments’ as the censoring reason are those patients who permanently withdrew from the study for various reasons (for example, adverse event, consent withdrawn) prior to the first scheduled tumour assessment, and their PFS times is censored at Day 1. Patients classified as ‘No on-study disease assessments’ in both PFS analyses (investigator assessment and BICR) had their PFS times all censored at Day 1. In the FDA analysis, these patients were classified with the specific reason for withdrawal as described in Tables 3 and 4 below.

In both the sponsor and FDA analyses, patients were censored at Day 1, and more importantly, both analyses showed the same number of patients with PFS events and censoring; therefore, any differences between the sponsor and FDA classifications for the reasons for censoring had no impact on the PFS analyses and the results were consistent.

Evaluator comment:


The sponsor’s response is accepted.
      1. Efficacy clinical question 5


QUESTION 5

For Study 1003, the sponsor is requested to provide a breakdown for both the control and experimental arms of the numbers and percentage where BICR was performed prospectively versus retrospectively.

Sponsor Response


The Blinded Independent Central Review (BICR) evaluation was incorporated in the Protocol Amendment 6 (dated 08Nov2012) for Study A5481003 after discussion with the US FDA in September 2012. At this time, the protocol had fully accrued all 165 randomised patients. There were tumor assessment imaging scans collected in a retrospective manner upon the implementation of the protocol amendment and prospectively for the assessments performed after the implementation of the protocol amendment at each participating site. The breakdown of numbers and percentage of the visits/time points for which BICR was performed retrospectively (on or prior to 08Nov2012) vs. prospectively (after 08Nov2012) are summarised below:

Table 37: Numbers and percentage of the visits/time points for which BICR was performed retrospectively (on or prior to 08Nov2012) vs. prospectively (after 08Nov2012)


Evaluator comment:


The reason for asking this and the following clinical question was that the first round evaluator identified investigator assessment as a significant potential source of bias in an open-label study such as this, and BICR was not undertaken from the outset of the study.

This response shows that there was a slightly higher rate of retrospective BICR sampling in the control arm. Whether this is of any significance is not clear, as one would expect BICR to remain unbiased whether images were collected before or after the decision was made that these should be reviewed centrally, as it is not in the taking of the image but in its review that the bias might be expected to manifest.

The sponsor’s response is accepted.

      1. Efficacy clinical question 6


QUESTION 6

For Study 1003, the sponsor is requested to provide concordance rates between the investigator and BICR by imaging modality eg bonescan, CT, MRI for each lesion type that is, bone lesions, visceral, other.

Sponsor Response


A patient can have more than one lesion types and multiple lesions of the same type. The investigator and BICR may observe different lesion types for a patient. Even if they observed the same lesion type, they may look at the different lesions.

The concordance rates between the investigator and BICR for bone lesions, visceral, other are 88.7%, 76.1%, and 81.1%, respectively.

The concordance rate is defined as

The total number of patients with the same lesion type either observed or not observed by both investigator and BICR divided by the total number of patients who had tumour assessment by both investigator and BICR


Evaluator comment:


The reason for asking this and the previous clinical question was that the first round evaluator identified investigator assessment as a significant potential source of bias in an open-label study such as this.

This response shows the rates of concordance were consistently between 75 and 90%. These rates are not unexpected, given the fraction of cases requiring adjudication for the central reviewers (that is, the rate of discordance between blinded reviewers) has been shown to be around 40%.17



The sponsor’s response is accepted.
      1. Efficacy clinical question 7


QUESTION 7

For Study A5481008, the sponsor is requested to state whether the CRF or Impala data was used to define populations for the primary efficacy analysis of the data, and how such discrepancies are handled in the statistical analysis. Noting that the 2016 ASCO meeting presentation of the latest results for Study 1008 (including BICR-reviewed data) by Dr Richard Finn used the CRF data, the sponsor is requested to provide a comment on the choice of this dataset over that presented for the primary analyses in this top-line summary.

Sponsor Response


Results provided in the top-line summary previously submitted in the registration application were stratified based on data from the study interactive randomisation technology (called ‘IMPALA’). As the randomisation in IMPALA occurred prior to the data being source verified by the sponsor, there were minor discrepancies in the distribution of patients by the stratification factors following data cleaning activities based on the CRF data. For this reason, the CRF source document verified data were considered more accurate and were used in all subsequent submission reports. The distribution of patients by the 3 stratification factors assessed based on investigator chosen-strata reported in IMPALA as well as derived from the data entered on the appropriate CRFs are summarised in Table 1 below.

As specified in the statistical analysis plan, the primary PFS analysis was performed using Disease Site (Visceral, non-visceral) recorded on IMPALA as the stratification factor to estimate the Hazard Ratio and to calculate the p-value. While Disease Free Interval and Prior Hormonal Therapy were also stratification factors for patient randomisation, they were not used for the primary PFS analysis.

A sensitivity analysis was performed to evaluate whether there was a difference when using CRF data for Disease Site as the stratification factor (see sensitivity analysis #4 in the Study A5481008 CSR provided). For sensitivity analysis #4, the results showed a HR of 0.572 with 95% CI (0.459, 0.713) similar to the primary PFS analysis which showed a HR of 0.576 with 95% CI (0.463, 0.718).

In conclusion, the minor discrepancies between the IMPALA data and CRF data on stratification factors did not have an impact to the primary PFS analysis.

Evaluator comment:


The sponsor’s response is accepted.
      1. Efficacy clinical question 8


QUESTION 8

For Study A5481008, what subgroups were prespecified for efficacy analyses?

Sponsor Response


The Study A5481008 Statistical Analysis Plan (SAP) version 2.0 was submitted in the initial registration application in Module 5.3.5.1 for Study A5481008. Version 3.0 was updated for study Protocol amendment 7 and is provided herein.

The SAP specified that the potential influences of baseline patient characteristics such as age, ethnic origin, ECOG performance status, geographical region, selected biomarkers, and stratification factors on the primary PFS, OS, and OR endpoints may be evaluated. In addition, to assess the impact of the concomitant use of proton pump inhibitors and the impact of palbociclib administered under fasting conditions, PFS was also evaluated in the population excluding patients who took proton-pump inhibitors and/or any other antacid medications concomitantly with study drug under fasting conditions during the active treatment phase.

Evaluator comment:


The sponsor’s response is accepted.
      1. Efficacy clinical question 9


QUESTION 9

How many patients ‘in follow up for progression’ in Study A5481008 were still on study drug in each arm?

The first round clinical evaluator comments relevant to this question were:



The censoring rates indicate similar absolute differences in discontinuations without disease progression across both arms. It is unclear how many patients ‘in follow up for progression’ were still on study drug (Clinical Questions) as this could include some where progression had not been established radiologically yet.

Clinical worsening in the absence of objective radiologically confirmed progression was handled differently in this study compared with Study 1003 and was required to be declared as PD only once there was radiological confirmation; clinical progression and discontinuation due to suspected progress was to be declared due to global deterioration and the numbers are even in both arms

Sponsor Response


As of the data cutoff date 26 February 2016 a total of 257 patients were still on study drug: 199 patients in the palbociclib + letrozole arm and 58 in the placebo + letrozole arm.

Evaluator comment:


This question pertains to one of the categories which tabulates the reasons for censorship in each arm, as it relates to the Kaplan-Meier curve of PFS in Study 1008.

The numbers cited by the sponsor are identical to the totals cited [table not in this report] under the category ‘in follow up for progression’. Therefore it appears that this is an error of communication, whereby the first round evaluator thought that ‘in follow up for progression’ referred to subjects in whom progression had occurred, and they were being followed up regardless. However, it appears the meaning of this category is actually ‘in follow up/not yet progressed’ that is, waiting to see progression which had not yet occurred.

This meaning should be confirmed by the formal review of Study 108. See section 13.

      1. Efficacy clinical question 10


QUESTION 10

When submitting the CSR for Study A5481008, the sponsor is requested to provide details of the deaths of each patient who died without evidence of disease progression if not already included in that document. This is recommended to be done as a second NCE application.

Sponsor Response


Study A5481008 Clinical Study Report (CSR) is provided to address the request for details of the deaths of patients who died without evidence of disease progression.

Deaths on study (that is, on treatment, included up to 28 days after the last dose of blinded therapy) as well as deaths that occurred during the follow-up period (that is, more than 28 days after the last dose of blinded therapy) as of the data cutoff date of 26 February 2016, are reported in Section 12.3.1.1 Study A5481008 CSR.

There were 7 (1.6%) patients in the palbociclib plus letrozole group and 2 (0.9%) patients in the placebo plus letrozole group who died on study without evidence of disease progression and were considered by the Investigator to be due to other/unknown causes (Table 49 Study A5481008 CSR).

During the post-treatment follow-up period, 7 (1.6%) patients died in the palbociclib plus letrozole arm for reasons other than progression and 2 (0.9%) patients in the placebo plus letrozole arm (Table 49 Study A5481008 CSR). Details and causality of deaths that were not due to disease progression are summarised in Tables 50 and 51 Study A5481008 CSR.

CIOMS narratives for these patients are provided in Section 14.3.3.1 Study A5481008 CSR.

Of note, none of the deaths without documented progression occurring in the palbociclib plus letrozole arm either on study or during the follow-up period was considered to be treatment – related by the Investigator (Table 50 Study A5481008 CSR).

Evaluator comment:


The sponsor indicates that the details of the deaths of each patient who died without evidence of disease progression have been included in the CSR document for Study 1008.

The sponsor’s response is accepted.


      1. Efficacy clinical question 11


QUESTION 11

Uncertainty exists as to whether there is a benefit for those with de novo metastatic disease. Whether this represents an increased responsiveness to the control arm which generally did better than in other subgroup analyses cannot be checked against the group who had received no prior systemic treatment for their disease (irrespective of stage of presentation) as these data could not be located in the Tables or Top-line summary. Provision of these data is requested to be included when the CSR is lodged with the TGA (second NCE application recommended).

Sponsor Response


An updated forest plot of subgroup analyses of PFS based on the CRF data is provided in Figure 1 for Study A5481008.

Prolongation of PFS in the palbociclib plus letrozole arm was demonstrated in all prespecified subgroups based on the stratification factors derived from the data recorded on the CRFs and baseline characteristics (Figure 1).

A reduction in the risk of disease progression or death in the palbociclib plus letrozole arm compared with the placebo plus letrozole arm was observed in all individual patient subgroups regardless of age, race, prior treatment, length of disease-free interval (see below for additional details), type of disease, and ECOG performance status at baseline.

This reduction in the risk was evident for all 3 disease-free interval subgroups: de novo metastatic disease (HR of 0.674 [95% CI: 0.457, 0.993]), disease-free interval ≤12 months since completion of prior (neo) adjuvant therapy (HR of 0.501 [95% CI: 0.329, 0.761]), and disease-free interval >12 months (HR of 0.516 [95% CI: 0.365, 0.731]).

Of note, using the data from the study interactive randomisation technology as previously provided in Figure 3 Study A5481008 Top Line Summary (submitted in the initial registration application), the upper bound of the 95% CI for patients with de novo metastatic disease was slightly above 1.0.

However, using the more accurate CRF-derived data (see Efficacy Question 7), the upper bound of the 95% confidence interval of HR for patients with de novo metastatic disease was <1.0 (Figure 1). Therefore the results from all 3 disease-free interval subgroup analyses were statistically significant.

The sponsor reviewed a number of the potential factors that may have influenced the HR estimate of the subgroup of patients with de novo metastatic disease. A comparison of patient key demographic and baseline disease characteristics across the disease-free interval (DFI) patient subgroups are summarised in Table 2.

All baseline patient demographic and disease characteristics are generally well-balanced between both treatment arms for each of the DFI subgroup. One notable exception is the proportion of patients with and without measurable disease in the de novo metastatic disease subgroup compared to the other 2 DFI subgroups. Patients with measurable disease represented approximately 90% of the de novo metastatic disease subgroup but approximately 70% in each of the other 2 DFI subgroups (Table 2). While a reduction in the risk of disease progression or death in the palbociclib plus letrozole arm compared with the placebo plus letrozole arm was observed regardless of the presence of measurable disease or not, the observed HR was higher for patients with measurable disease (HR of 0.663 [95% CI: 0.517, 0.849]) than for patients with no measurable disease (HR of 0.350 [95% CI: 0.215, 0.568]) at baseline. This observed HR for patients with measurable disease was similar to that observed for patients with de novo metastases of which 90% had measurable disease. Additionally, 151 out of the 157 patients with no measurable disease were patients with bone-only disease for whom a statistically significant reduction in the risk of disease progression or death in the palbociclib plus letrozole arm compared with the placebo plus letrozole arm was observed (HR of 0.363 [95% CI: 0.221, 0.594]) (Figure 1).

Additionally, as of the data cut off (26 Feb 2016), in the de novo metastatic subgroup the percentage of patients with events in the palbociclib plus letrozole arm versus placebo plus letrozole was 39.5% vs.51.9%, respectively. By comparison, the percentage of patients with events in each treatment arms in the ≤ 12 months and > 12 months DFI subgroups were 53.6% versus 77.1% and 40.4% versus 61.3 %, respectively. It is important to note that a smaller percentage of patients with de novo metastatic disease had a PFS event in the placebo plus letrozole arm (51.9%) compared to the other two DFI subgroups where greater percentages of patients with events were observed (77.1% and 61.3%, respectively) suggesting that placebo plus letrozole data for the de novo metastatic subgroup may not be as mature as the data from the other two DFI subgroups to determine the median PFS for patients with de novo metastatic disease. As there were patients still on treatment at the time of data cutoff, additional PFS events will occur and thus, a future PFS analysis may provide further confirmation of the clinical benefit of palbociclib plus letrozole in this subgroup of patients.

Results of the prespecified subgroups analyses based on stratification factors derived from the data recorded on the CRFs and baseline demographics and disease characteristics, supporting the consistency of PFS benefit findings within the study population are detailed in the A5481008 CSR.

In conclusion, the higher percentage of patients with measurable disease in the de novo subgroup and the maturity of the data at the time of the data cutoff, offer possible explanations for the higher observed HR for this subgroup than that observed for the other 2 DFI subgroups.

Evaluator comment:


The HR for the subgroup of patients with metastatic disease at diagnosis (de novo) for the active versus placebo arm falls just inside nominal statistical significance (0.674 [95% CI: 0.457, 0.993]), when the stratification is performed using data that has been cleaned based on CRF information rather than unaltered data from an interactive response technology (IRT) system ‘IMPALA’ (see Question 7 and response). With the IMPALA-based stratification, the result was HR 0.729 (95% CI: 0.486, 1.093; p=0.063).

In terms of the meaning of this for statistical significance, even with the cleaned data this result is only just borderline significant.

However, it is agreed that the rate of PFS events in the placebo arm for the de novo metastatic subgroup is lower than those in the placebo arms whose disease was recurrent after remission (for whatever duration). It is possible that the data is not yet fully mature and additional evidence of PFS benefit in the de novo group may accrue with further follow up.

The relevance of measurable disease and bone-only disease are not clear from this analysis – these may be confounders if on the same causal pathway as de novo disease, but that subgroup analyses for them were significant does not change the borderline significance of the de novo subgroup analysis.

Whilst there is not strong evidence that palbociclib adds to efficacy of letrozole in de novo metastatic disease, there is some, and there is no biological reason to suspect that de novo metastatic disease should be expected to respond any differently to those in the other categories of duration since remission.

Given that the data cut-off date was ten months ago, a reanalysis for efficacy in the de novo subgroup may be valuable but is not necessary to support registration. The submission of further data to this point when available could be made a condition of registration.


      1. Efficacy clinical question 12


QUESTION 12

For Study 1023, the anticipated dropout rate was high at 25%, particularly for those with metastatic disease and a high degree of motivation to continue treatment if their disease is not progressing. The sponsor is requested to provide a rationale for this high rate. Was this to anticipate side effects related to the use of fulvestrant, the administration of which is associated with significant discomfort?

Sponsor Response


The 25% anticipated dropout rate was an estimation and was only used at the study design stage for planning purposes. For a given Hazard Ratio (HR) of PFS to be detected in the study, a higher dropout rate would not increase the number of PFS events but would extend the follow-up time for cumulating events. This would result in the enrollment of more patients in order to shorten the follow-up time. In fact, for the analysis of PFS only about 11% of patients’ had PFS times that were censored for various reasons that could be considered as drop out (see yellow highlighted rows in Table 1 below). This rate was much lower than the estimated rate of 25% used for planning purposes.

Evaluator comment:


The sponsor’s response is accepted.
      1. Efficacy clinical question 13


QUESTION 13

In the update report using an earlier cut-off date of 16 March 2015, discontinuations due to withdrawal of consent occurred in 1.2 % (4 patients) in the palbociclib and fulvestrant arm but are now reported as 0.9% (3 patients) with a later cut-off date - the sponsor is requested to explain why there are now fewer presented (Clinical Questions).

Sponsor Response


The review and cleaning of data is an ongoing process which is active until all the patients complete their treatment and their survival status is reported. In particular, at the time of the data cutoff date of 05 December 2014 which was used for data reporting in the CSR, most patients were still on treatment and site visits for data review were conducted very frequently. Therefore, some data modifications may have occurred due to query resolution, source data verification or additional information that may become available to the investigators after the data snapshot for data reporting.

During subsequent data reviews, it was identified that the investigator reported ‘withdrew consent’ as reason for treatment discontinuation for Patient [information redacted] (Table 1023.683.1; data cutoff date 16 March 2015), but this patient refused to continue treatment and accepted to proceed with the follow-up visits, thus making the reason for treatment discontinuation (that is, ‘withdrew consent’) inappropriate. A query was raised and the investigator changed the reason for treatment discontinuation to ‘subject refused continued treatment for reason other than adverse event’, which was the reason for treatment discontinuation in the clinical database as of the data cutoff date of 23 October 2015 (Tables 1023.683.2 and 1023.683.3).

Evaluator comment:


The sponsor’s response is accepted.
      1. Efficacy clinical question 14


QUESTION 14

The sponsor included an updated PFS analysis for Study 1023, which also included updated OR, DoR and clinical benefit rate. None of these analyses were supported by BICR-derived analyses; the sponsor is requested to provide the BICR analyses for these endpoints for evaluation or state that none was done.

Sponsor Response


In Study A5481023 a sample-based Blinded Independent Central Review (BICR) approach was to be implemented as an auditing tool for PFS as per the protocol and the Statistical Analysis Plan (SAP). The objective of this approach was to corroborate the results of the primary investigator-assessed PFS analysis and to evaluate for any potential bias. The BICR audit approach was not intended to provide an alternative means of definitive analysis. Of note, this approach was discussed with and agreed upon by both the US FDA and EU EMA.

The External-Data Monitoring Committee met on April 2015 to review the results of the prespecified interim analysis (data cutoff date 05 December 2014) and concluded that the study met its primary objective of prolonging the investigator-assessed PFS in the palbociclib plus fulvestrant arm compared to the placebo plus fulvestrant arm. This interim analysis was therefore considered the primary analysis of PFS and a PFS analysis based on BICR data from the randomly sampled audit subset (n=211) was then conducted. Although not required by protocol and SAP, the PFS analysis based on the BICR audit approach was again performed at the time of the first PFS update (data cutoff date 16 March 2015) to further corroborate the investigator–assessed PFS analysis on a more mature dataset. The results of updated PFS analysis based on BICR audit approach (n=211) at the data cutoff date of 16 March 2015 are presented in Table 1 and the results of secondary efficacy analyses of Objective Response and Clinical Benefit Rate in Table 2 and Duration of Response in Table 3.

The results of the updated PFS analysis based on BICR audit approach as of the data cutoff date of 16 March 2015 were consistent with the primary and updated analyses of the ITT population and with the BICR data at the data cutoff date of 05 December 2014. The HR was 0.375 (95% CI: 0.233, 0.604; stratified 1-sided p-value 0.000015) in favor of palbociclib plus fulvestrant. The median PFS was not reached (NR) (95% CI: 10.9, NR) for the palbociclib plus fulvestrant arm and was 3.8 months (95% CI: 3.4, 9.3) for the placebo plus fulvestrant arm (Table 1023.412.3).

Also, based on data as of the cutoff date of 16 March 2015, the differential discordance, as determined by the Early and Late Disagreement Rates (20.05% and -21.43%, respectively), was not suggestive of any investigator bias in favor of the palbociclib plus fulvestrant arm (Table 1023.412.9).

Based on the above findings which confirmed the consistency between the investigator-assessed PFS with the BICR-based audit approach, an additional BICR assessment of the second updated PFS analysis from the 23 October 2015 cutoff date was not considered necessary by the sponsor.

Evaluator comment:


The sponsor’s response is accepted.
      1. Efficacy clinical question 15


QUESTION 15

Table 16, CSR for Study A5481023 includes data about the recurrence type. This includes ‘newly diagnosed’ as a significant category (17.7% of total population) amongst breakdown by anatomical site which makes it difficult to establish how many in the study had local or locoregional disease only. The sponsor is requested to provide:

  1. A breakdown of numbers in each arm this is a population identified in the indication;

  2. The following efficacy outcomes for those in each arm with local or locoregional disease: median PFS, OR, DoR, CBR.

Sponsor Response


The sponsor has provided a listing reporting sites of disease for each patient categorized as ‘newly diagnosed’ by the investigators (Table 1023.679.1).

Most patients in both the palbociclib plus fulvestrant arm and in the placebo plus fulvestrant arm had a distant recurrence (for example, bone involvement in the disease). One patient in the palbociclib plus fulvestrant arm was found to meet the definition of ‘locoregional recurrence’ ([information redacted]). This patient had a recurrence in the chest wall with only 2 superficial lesions that were followed during the study through physical examinations. This patient was added to the other patients with Local recurrence and Locoregional recurrence to provide the requested efficacy outcomes of the subset of patients with local/locoregional recurrence.

The analyses presented in this response were conducted on the dataset corresponding to the data cutoff date of 23 October 2015. Some minor differences in the number of patients with local/locoregional recurrence reported in the CSR (data cutoff date 05 December 2014) and those reported in the present analyses (data cutoff date 23 October 2015) were due to the data review and cleaning process. This process is ongoing because there are patients in the study who are still on treatment.

Progression-free survival (PFS) of patients with local/locoregional recurrence is presented in Table 1, and objective response rate (ORR), duration of response (DoR) and clinical benefit response rate (CBRR) in Table 2.

As shown in Table 1 the benefit of palbociclib plus fulvestrant versus placebo plus fulvestrant in significantly prolonging PFS is also apparent in patients with local/locoregional recurrence. A numerically higher ORR which did not reach statistical significance was demonstrated for palbociclib plus fulvestrant versus placebo plus fulvestrant (Table 2). This is very likely due to the small number of patients with tumor response in the subset of patients with local/locoregional recurrence. Median DoR was similar for the 10 patients in the palbociclib plus fulvestrant arm and the 3 patients in the placebo plus fulvestrant arm who had a tumor response. The analysis of CBR (CR, PR and SD≥ 24 weeks) in the subset of patients with local/locoregional recurrence demonstrated a numerically higher CBR for palbociclib plus fulvestrant compared with placebo plus fulvestrant (Table 2), which did not reach statistical significance.

In conclusion, a clinically and statistically significant improvement in PFS of palbociclib plus fulvestrant over placebo plus fulvestrant was observed also in the subset of patients with local/locoregional recurrence, benefit that was also maintained in ORR and CBR.

Evaluator comment:


Table 1023.679.1 is not reproduced in this CER but has been reviewed. As described by the sponsor, there was only one subject in this listing (subject [information redacted]) who had only local disease recurrence.

The sponsor’s response is accepted.


      1. Efficacy clinical question 16


QUESTION 16

This question only needs to be addressed in the s31 response if the sponsor wishes to retain the PI statements about quality of life in the Clinical trials section:

  1. a justification of the clinical significance of the presented results against the prespecified criteria in the SAP;

  2. to indicate meaningful completion rates, please provide the number and percentage of patients who completed all questions of the EORTC-QLQ-C30 out of the PRO analysis in each arm;

  3. please provide the number / percentage of patients for whom pro-rating was undertaken due to missing data in each arm.

Sponsor Response


The sponsor has acknowledged the TGA comments and agreed to remove the quality of life claims made based on Global Health Status/QoL and emotional functioning, requested in PI Question 19.

The sponsor proposes to replace the above mentioned information with a prespecified time to event analysis for deterioration in pain (TTD), defined as first occurrence of an increase of at least 10 points in pain symptom scores.. Statistically convincing and plausible results were achieved with a difference in median time to deterioration of 8.0 versus 2.8 months, HR 0.6, p <0.001.

The TTD statement proposed in the PI aligns with the approved SmPC. The sponsor considers this useful information for prescribers and hence proposes it be included in the Product Information:

Time to Deterioration (TTD) was prespecified as time between baseline and first occurrence of ≥ 10-point increase from baseline in pain symptom scores. Addition of palbociclib to fulvestrant resulted in a symptom benefit by significantly delaying TTD in pain symptom scores compared with placebo plus fulvestrant (median 8.0 months versus 2.8 months; HR of 0.64 [95% CI: 0.49, 0.85]; p<0.001).’


Evaluator comment:


The evaluator referred to the answer to a question raised regarding the PI.18
      1. Efficacy clinical question 17


QUESTION 17

In Study A5481023, no information was found by the evaluator as to how many of the biopsy samples used in the central testing were from a biopsy sample taken following their most recent episode of progression to determine ER/PR/HER2 status – the sponsor is requested to provide this information as it has been shown that a discordant rate between primary and secondary breast cancers has been reported to be as high as 25-30%.

Sponsor Response


Study A5481023 required eligible patients to provide a tissue biopsy sample taken at the time of presentation with recurrent or metastatic disease. A de novo biopsy was required only if no archived tissue of metastatic disease was available at study initiation. Archival tissue of primary disease was acceptable only for patients with bone only disease and for those patients who entered the study just after disease progression while receiving adjuvant therapy.

The protocol did not require the collection of the exact tumor biopsy site information including whether it was a recent metastatic site or not.

The time between the date of patient tumor biopsy and the date of randomisation was less than 1 month for 19% of tissue samples, between 1 and 2 months for 11% of samples, and between 2 and 6 months for 8% of samples. Twelve percent of tissue samples were biopsied more than 6 months and within 12 months before randomisation. There were 50% tissue samples biopsied more than 12 months before randomisation.

The sponsor would like to comment that while it is recognized that discordance in human epidermal growth factor receptor 2 (HER2) status and hormonal receptors status between the primary breast tumor and the corresponding metastatic lesion is frequent, mainly when adjuvant chemotherapy was the treatment of choice, it is not demonstrated that additional modifications of hormonal and HER2 receptors status occur when the disease is already metastatic (1).

Current best practice is to check receptor status from the most accessible metastatic lesion before selecting the treatment for a patient with metastatic disease. For these reasons, and also considering the objective difficulties to biopsy all the metastatic lesions of a patient, the sponsor adopted the current best practice to rigorously select patients to be eligible for the study.

References

1) Curtit E, Nerich V, Mansi, L et al. Discordances in Estrogen Receptor Status, Progesterone Receptor Status, and HER2 Status Between Primary Breast Cancer and Metastasis. Oncologist. 2013 Jun;18(6): 667–674

Evaluator comment:


  • Site of biopsy for the sample that was used to identify hormone receptor status was not required to be recorded per protocol.

  • Proportion of biopsies that were de novo at study entry versus archived metastatic tissue versus archived primary tissue has not been provided.

  • In Study 1003, a large proportion of subjects (22.7% of the ITT population) had bone-only disease at diagnosis of metastatic disease.

  • For half of the samples the biopsy that was used was at least a year old prior to randomisation, suggesting a high proportion of samples were archival tissue. Whether these were of secondary or primary disease is not clear.

  • On the subject of discordance:

    • A 2016 study of tumour expression profiles found ‘primary tumors and metastases were highly concordant for HER2 (84 %, p = 1.13E-08), ER (90 %, p = 3.26E-10) and PR (83 %, p = 2.09E-09) and ER-and PR-positive metastases were significantly found to be of visceral origin (p = 0.03, p = 0.02).’19

    • A literature review on the topic in 2015 found that PR varied more than ER, at a rate of 10% to 30% for ER and 20% to 50% for PR.20 Loss of PR co-expression, that is, ER-positive/PR- is often seen in subjects who have developed resistance to an endocrine agent like tamoxifen.21

    • A 2014 meta-analysis notes that ‘clinical management of breast cancer metastasis has been based largely on the initial assessment of the primary tumor.’22 In this study, a large proportion of discordance in hormone receptor status was attributed to the ‘limited accuracy of receptor assays’. They state ‘The corrected discordance in ER between primary tumors and recurrent or metastatic lesions was 12.4%, and there were more positive-to-negative changes (10.1%) than negative-to-positive changes (2.3%). Similar patterns were observed for progesterone receptor (PR), although the overall discordance in PR was higher.’

  • In the worst case scenario, if all of the tumour samples that were at least a year old by time of randomisation (~50%) were archival primary tissue, then it is likely that 5% to 15% of the total study population could have had discordant HR profiling.

  • Subjects more likely to have discrepancy would be those who’d had prior chemotherapy or endocrine therapy.

  • The ASCO guidelines on the use of biomarkers to guide decisions on systemic therapy for women with metastatic breast cancer (published April 2015)23 state that ‘if discordances are found, evidence is lacking to determine whether outcomes are better with treatment regimens based on receptor status in the metastases or the primary tumor.’

  • The biopsy sources in Study 1008 likely to be reflective of real-world practice and give reasonable external validity to the findings. It is also more likely that changes in receptor status went from HR+ to HR- than the other way around, making a type 2 error more likely than a type 1 error, so it is unlikely that this has led to an overestimation of effect.
Conclusions:

It is unlikely that then lack of certainty around origin of biopsy has significantly biased the efficacy results of this study, although it does introduce some additional uncertainty.

The sponsor’s response is accepted.


      1. Efficacy clinical question 18


QUESTION 18

Study 1023: The sponsor is requested to provide 3 additional sensitivity analyses, presented as a forest plot with accompanying HRs comparing the ITT PFS:

    1. A sensitivity analysis, removing all those who were ineligible for enrolment due to subsequently determined ER-/PR- or HER2+ disease by central testing, to determine whether there was any effect on the ITT PFS analysis;

    2. A second sensitivity analysis excluding those 118 patients whose data were missing or inadequate for central laboratory testing of ER,PR or HER2 status as well as those who were deemed ineligible by central laboratory testing, is also requested to determine whether there was any effect on the ITT PFS analysis;

    3. A subgroup analysis of those whose results were discordant, that is not ER-positive or was HER2-positive.

Sponsor Response


As requested, the sponsor provides the forest plot of the primary Progression-free survival (PFS) and 3 sensitivity analyses (Figure 13) based on the data cutoff date of 23 October 2015.

Sensitivity Analysis a. Excluding patients who do not meet the criteria of HR+ (ER-positive or PR-positive) and HER2- categorized by central laboratory.

Sensitivity Analysis b. Excluding patients who do not meet the criteria of HR+ (ER-positive or PR-positive) and HER2- categorized by central laboratory and patients whose data were missing or inadequate for central laboratory testing of ER,PR or HER2 status.

Sensitivity Analysis c. Patients with at least one discordant ER, PR, and HER2 status between local and central laboratories.

Figure 13: Forest plot of the primary and sensitivity analyses



Summaries of the 3 sensitivity analyses are provided in Tables 1 to 3.

In conclusion, the sensitivity analyses associated with exclusion of patients with differences in ER/PR/HER2 receptor status between local laboratories and central laboratory confirmed the benefit of palbociclib plus fulvestrant versus placebo plus fulvestrant in significantly prolonging PFS. The results of these sensitivity analyses further support the robustness of the primary PFS analysis.

Evaluator comment:


The sponsor’s response is accepted.
      1. Efficacy clinical question 19


QUESTION 19

The sponsor is requested to provide the number of patients in each arm whose tumour was ER-/PR-positive and the results (PFS, OR, DoR, and CBR) for these patients.

The round 1 clinical evaluator comments relevant to this question were:

The sponsor has also been requested to provide a breakdown of the numbers and outcomes for those with ER-/PR-positive disease as this population is currently encompassed by the proposed indication, but no efficacy outcomes are provided specifically for this group. Until this information is provided, any potential recommendation may require that the indication is restricted to those with ER-positive disease only.

Sponsor Response


Study entry testing for eligibility criteria of ER, PR, and HER2 was based on local laboratory results utilising assays consistent with local standards for the reporting of the efficacy endpoints. Central assessment of ER, PR, and HER2 on tumor samples with adequate quality and quantity were performed retrospectively at the central laboratory Clairent diagnostic services, Inc. The efficacy results for patients who were identified as ER-/PR-positive by local testing is provided in Table 1 and by central laboratory testing in Table 2.

Evaluator comment:


In this study (Study 1023), there were 3 subjects by local laboratory testing and 13 subjects by central laboratory testing who had ER-/PR-positive baseline disease. Only one of these subjects had an objective response, which lasted 3.71 months. This subject was in the placebo arm. This trial (Study 1023) does not provide any evidence of efficacy in subjects with ER-negative/PR-positive disease however the very small population in whom ER was negative precludes drawing statistically significant conclusions. Study 1003 had ER positivity as an inclusion criterion, and so wouldn’t include any ER-/PR-positive disease.

On this basis alone, it may appear that restriction of the indication to ‘ER-positive’ rather than ‘HR+’ disease would be appropriate if there is no evidence of efficacy in ER-negative/PR-positive disease seen Study 1008. However, the following information suggests that this might not be appropriate:



  • The PR is an estrogen-regulated gene – so PR activity is modulated by ER activity rather than the other way around.24

  • PR positivity in the absence of ER positivity is rare – approximately 2% of breast cancers,25 and these patients tend to be treated the same as ER-positive/PR-positive patients,26 that is, the NCCN guidelines refer only to ‘hormone receptor’ positive or negative, not specifying ER/PR status separately.27

  • Receptor status can change during the course of disease and at any one time may not be uniform throughout the body as tumour clonal lines continue to evolve, so receptor status may not be consistent between one biopsy and another. This is more common for PR (20-50% differences between locoregional and distant metastases) than for ER (10-30%), whilst on the contrary, HER2 status tends to show high concordance.20

  • Some authors argue that ER-/PR-positive breast cancer is not a reproducible subtype, and the value of PR testing is questionable, with no association seen between PR status and prognostic value in multivariate modelling.28

  • There appears to be poor concordance of testing between peripheral and central laboratories as evidenced by the complete disagreement between central and local laboratory results in this group (there were zero subjects identified as ER-/PR-positive by both laboratories in agreement). This supports that there is variability between laboratory testing sites as well as between biopsies or due to clonal evolution with time.

  • ER-/PR-positive/HER- patients tend to have more aggressive tumours (hypothesised to have similar to clinical outcomes in triple negative patients),26 so the difference between the efficacy rate in ER-positive/PR-positive and ER-/PR-positive subjects in this study may reflect the natural history of the disease rather than lack of efficacy in the latter group.

  • The sponsor’s response is accepted.
      1. Efficacy clinical question 20


QUESTION 20

Study A5481023: The change to a different schedule (2 weeks on/2 weeks off) was not described in the Protocol and the benefits of this dose in the 3.8% who switched to this regimen cannot be assured. It is not clear what palbociclib dose was taken in this regimen (please provide this information). It would be apparent to the investigator that the patient was receiving palbociclib due to the AE profile, and this may have introduced a bias in wanting to continue if there was a clinical benefit observed but problematic toxicities. That so many changed to this regimen indicates a degree of unmasking. The sponsor is requested to provide an explanation of whether these non prespecified alterations were included in the protocol deviations and also whether these patients’ outcomes were included in the efficacy analyses, and whether any subgroup efficacy analyses were performed for those on this regimen.

Sponsor Response


The sponsor would like to clarify that the palbociclib dose 75 mg/day schedule 2 weeks on/2 weeks off is allowed by the Protocol of Study A5481023. In protocol Section 5.3.4.2.3 (Dose Reductions), Table 3 reports the guidelines to modify the dose of palbociclib/placebo when treatment-related toxicities occurred. Table 3 with related footnotes is provided in this document.

In Table 3 one of the footnotes indicates that when a patient was already receiving palbociclib at 75 mg/day (or placebo equivalent) according to schedule 3 weeks on/1 week off, it was possible to consider changing the schedule to 75 mg/day schedule 2 weeks on/2 weeks off.

In this clinical trial, eligible patients may have received several prior anti-tumor agents, including chemotherapy, thus increasing the propensity to develop toxicities during a subsequent anti-tumor therapy (that is, study treatment). For this reason, the protocol provided oncologists and patients with an additional possibility to adapt the palbociclib/placebo dosing in order to continue with the study therapy when an objective clinical benefit was observed (for example, tumor response or stabilization).

Study A5481023 results demonstrated that adding palbociclib to fulvestrant a clinically meaningful and statistically significant benefit in Progression-free survival (PFS) was obtained for patients with estrogen receptor -positive, human epidermal growth factor receptor 2 (HER2)-negative advanced breast cancer.

At the data cutoff date of 23 October 2016, a total of 14 patients in the palbociclib plus fulvestrant arm had the palbociclib dose reduced to 75 mg/day schedule 2 weeks on/2 weeks off (Table 1023.565.14). Of these patients, 5 were still ongoing as of 31 October 2016 with >2 years of treatment and 9 patients had discontinued study treatment, most of whom were due to disease progression. Overall, the 75 mg/day schedule 2 weeks on/2 weeks off dosing regimen was administered for <6 months in 5 patients, for approximately 10 months in 3 patients and for ≥ 1 year in 6 patients.

A specific PFS analysis of those patients who received 75 mg/day schedule 2 weeks on/2 weeks off was not conducted.

However, the PFS analysis of the subgroup of patients who had the dose of palbociclib decreased to 75 mg/day, which includes also patients with a dose decreased to 75 mg/day schedule 2 weeks on/2 weeks off, was conducted and is provided in Table 1 together with the PFS analysis of patients who had at least one palbociclib dose reduction and those who had a dose reduction to 100 mg/day. As shown in Table 1, the PFS of the 3 subgroups of patients is comparable.

In conclusion, the sponsor considers that the possibility of continuing treatment, although at a reduced total number of mg of palbociclib over a treatment cycle compared to the 3 weeks on/1 week off regimen, still represented a clinical benefit for patients in this setting.

Comment: The PFS analysis for the smaller dose subgroups given by the sponsor is outlined in Table 38 below, with updated PFS results in the ITT placebo and active groups, which was provided by the sponsor in the response) for comparison.

Table 38: PFS in reduced dose subgroups and ITT population, from Study 1023 (data cut-off: 23Oct2015)



Group

N =

Objective progression

Death without progression

K-M estimate of median PFS in months (9% CI)

Placebo + fulvestrant (ITT population of Study 1023)

174

130 (74.7%)

3 (1.7%)

4.6 (3.5, 5.6)

Palbociclib + fulvestrant (ITT population of Study 1023)

347

198 (57.1%)

2 (0.6%)

11.2 (9.5, 12.9)

Palbociclib + fulvestrant subgroup: patients with at least one dose reduction

132

66 (50.0%)

0

14.1 (11.1, 16.7)

Palbociclib + fulvestrant subgroup: Patients with dose reduction to 100 mg/day

89

46 (51.7%)

0

11.3 (9.2, N.E.)

Palbociclib + fulvestrant subgroup: Patients with dose reduction to 75 mg/day

43

20 (46.5%)

0

16.7 (11.2, N.E.)

The efficacy results in the subgroups with reduced doses are in keeping with those of the active arm generally, and do not overlap in confidence interval with the placebo arm. The dose reduction for management of adverse events was pre-specified, and blinding appears to have been maintained as dose reduction could have been undertaken for placebo also.

The sponsor’s response is accepted.


      1. Efficacy clinical question 21


Evaluator comment: this question is more related to safety than efficacy but the nomenclature of ‘efficacy’ question 21 is kept for continuity between documents.

QUESTION 21

Study A5481023 the sponsor is requested to state whether searching using any other MedDRA terms that might capture events of haemorrhage, bleeding or bruising, yields any events associated with the adverse events of thrombocytopenia.

Sponsor Response


The sponsor performed a search for the following Preferred Terms (PTs) that capture events of hemorrhage: Haemorrhage, Petechiae, Menorrhagia, Haematochezia, Haematuria, Epistaxis, Haemorrhoidal haemorrhage, Haemorrhage intracranial. A review of data from patients treated on the palbociclib and fulvestrant arm with adverse events of hemorrhage as captured by the above listed PTs and with laboratory findings of thrombocytopenia was performed.

Two cases of Grade 1 Epistaxis and Grade 1 Thrombocytopenia were identified. In one of the cases, the event Epistaxis did not have temporal plausible relationship with the event Thrombocytopenia, since the events occurred 3 months apart. In the other case, intermittent Grade 1 thrombocytopenia started in Cycle 2. The patient had 2 episodes of Epistaxis; the first occurred on Cycle 1 Day 22, lasted for one day and was assessed as related to study medication. No action was taken with study medication. The second event occurred on Cycle 5 Day 26; the event lasted one day and was considered unrelated to study medication but related to other illness. No action was taken with study medication. The patient was concomitantly taking ibuprofen.

Additionally, this search identified 35 adverse events (AE) reported in 31 patients. There were 29 cases of Epistaxis, 2 cases of Menorrhagia (2 events in 1 patient), 1 case of Haematochezia, 1 case of Haematuria, 2 cases Haemorrhoidal haemorrhage and no cases for the remaining PTs listed above.

Please see Table 39 for a listing of pertinent adverse events by treatment arm.

Table 39: Adverse events by treatment arm



An analysis of the platelet counts was also performed for all patients in Study 1023. This analysis focused on 3 cycles: the cycle in which the adverse event occurred, as well as the preceding and subsequent cycle.

The review also included an analysis of concomitant medications, specifically aspirin, NSAIDs, and COX-2 inhibitors. 12/31 of patients with bleeding events were taking one or more of these concomitant drugs.

A review of the 35 cases of AEs of hemorrhage identified only 1 case in which a laboratory abnormality was not reported as an AE. This was a Grade 2 decrease in platelets that occurred 6 days after a Grade 1 Epistaxis was reported. This patient was concomitantly taking aspirin.

It is the responsibility of the investigator to determine when a laboratory test abnormality is to be considered an AE. As discussed in Safety Question 10, during the data review/cleaning process queries were generated for laboratory test abnormalities that were not reported as AEs. In some instances, investigators did not consider laboratory test abnormalities with a severity of Grade ≤2 as medically relevant and thus these may not have been reported as an AE.

In conclusion, a search for the above specified PTs pertinent for bleeding events and concurrent laboratory tests indicative of thrombocytopenia identified only a single case in which there was an association between the event (Epistaxis) and thrombocytopenia.

Evaluator comment:


The sponsor’s response is accepted.
      1. Efficacy clinical question 22


QUESTION 22

Study A5481008: Within the study, there was some discordance between the baseline information provided at randomisation which affected the stratification and has had an impact on the efficacy analyses, particularly on the subgroup analyses, depending which dataset is used for the ITT population. The full impact of these cannot be understood and contextualised without presentation of the study protocol deviations. It is not sufficient to provide the analyses for these groups according to the differing information source (that is, randomisation versus CRF). A more rigorous approach should include:

  1. Presentation of the number of patients for whom there was any discordance between the randomisation information and CRF;

  2. Whether these patients were from a single or limited number of investigation centres - it is noted that in Study 1003, the FDA clinical site audit identified a single site as having a significant number of protocol deviations but that analyses with these patients censored were not reported to significantly affect the outcomes;

  3. Presentation of sensitivity analyses for the efficacy outcomes censoring the data from these patients incorrectly classified.

Sponsor Response


The sponsor refers the TGA to response to Efficacy Question 7 which clarified that because randomisation in IMPALA occurred prior to the CRF data being source verified, there were minor discrepancies in the distribution of patients by stratification factors for one or more of the investigator-chosen strata not matching the data recorded on the applicable CRFs. For this reason, the CRF source document verified data were considered more accurate and were used in all subsequent submission reports. The data based on the CRF were similar to those based on randomisation and the distribution of patients by the 3 stratification factors (Disease Site, Disease Free Interval, Prior Hormonal Therapy) was similar between the 2 treatment arms and did not have an impact on the primary PFS analysis.

Question a):

A summary of the concordance of patient by stratification factor is presented in Table 1 (disease site), Table 2 (disease free interval), and Table 3 (prior hormonal therapy). Disagreements between IMPALA and the CRF data were identified in 30 patients (23 in the palbociclib + letrozole arm and 17 in the placebo + letrozole arm) for the ‘disease site’ stratum, 59 patients (41 and 18, respectively) for the ‘disease free interval’ stratum and in 7 patients (4 and 3, respectively) for the ‘prior hormonal therapy’ stratum. Patients with more than one stratification disagreement are counted in each group. Geographic distribution of these disagreements and evaluation of their impact on the PFS primary analysis are summarised below in responses 22b and 22c, respectively.

Question b):

A total of 86 (12.9%) patients from 67 sites across 16 countries were randomised in IMPALA with one or more stratification factors not matching the data recorded on the applicable CRFs. A summary of the distribution of the stratification disagreements by country and by sites per treatment arm is provided in Table 4. Not one specific site was identified as an outlier with a large number of stratification-related protocol deviations.

Question c):

As clarified in Efficacy Question 7, while Disease site, Disease Free Interval and Prior Hormonal Therapy were all stratification factors for patient randomisation, the primary PFS analysis was performed using only Disease Site (Visceral, non-visceral) as the stratification factor to estimate the Hazard Ratio and to calculate the p-value. Consequently, the sponsor conducted a sensitivity analysis of PFS excluding only patients whose Disease Site (Visceral, non-Visceral) was assigned differently in IMPALA compared to the CRF data. A total of 30 patients (23 patients in the palbociclib-letrozole arm, 7 patients in the placebo-letrozole arm) were excluded from this sensitivity analysis.

A comparison of the PFS analysis results from the sensitivity analysis and the primary analysis is summarised in Table 5.

Results from the sensitivity analysis presented in Table 5, as well as from the sensitivity analysis #4 reported in the A5481008 CSR (summarised in Efficacy Question 7), demonstrate that the discrepancy between the IMPALA data and CRF data on the stratification factor used for the primary PFS analysis did not significantly impact the results of the primary PFS analysis.

Evaluator comment:


The largest discrepancies between the IMPALA and the CRF data appear to be in the disease-free interval assignment of category. The category of metastatic disease with onset longer than 12 months after adjuvant therapy finished had the highest rate of misclassified subjects, with redistribution of 7% of the active arm and 5.9% of the placebo arm from this category to another (de novo metastatic or up to 12 months post adjuvant therapy).

There was also a slightly higher percentage of the active arm misclassified as visceral versus nonvisceral (both directions), with total 5.2% of the active and 3.2% of the placebo arm changed category. In both arms, slightly more patients were reclassified to non-visceral from visceral.

Prior hormonal therapy differed the least of the three strata investigated for CRS/IMPALA concordance.

The overall disagreement rate was 13.5% in the active and 11.7% in the control arm, and no one site was a particular culprit for misclassification – instances were sparsely distributed amongst the study centres.

The primary PFS analysis was stratified only by disease site, not the other randomisation factors. A sensitivity analysis of the primary outcome excluding patients whose disease site had been reclassified (23 patients in the active and 7 in the comparator arm) yields a very similar result to the primary analysis (HR 0.572 versus 0.576).

The sponsor’s response is accepted.


1   ...   37   38   39   40   41   42   43   44   ...   53


The database is protected by copyright ©dentisty.org 2016
send message

    Main page