Original Article
`
`Patient-Reported Outcomes From EMILIA, a Randomized
`Phase 3 Study of Trastuzumab Emtansine (T-DM1) Versus
`Capecitabine and Lapatinib in Human Epidermal Growth
`Factor Receptor 2–Positive Locally Advanced or Metastatic
`Breast Cancer
`
`Manfred Welslau, MD1; Veronique Dieras, MD2; Joo-Hyuk Sohn, MD3; Sara A. Hurvitz, MD4; Deepa Lalla, PharmD, PhD5;
`Liang Fang, PhD5*; Betsy Althaus, PharmD, BCOP5; Ellie Guardino, MD, PhD5; and David Miles, MD6
`
`BACKGROUND: This report describes the results of an analysis of patient-reported outcomes from EMILIA (TDM4370g/BO21977),
`a randomized phase 3 study of the antibody–drug conjugate trastuzumab emtansine (T-DM1) versus capecitabine and lapatinib in
`human epidermal growth factor receptor 2 (HER2)–positive locally advanced or metastatic breast cancer. METHODS: A secondary
`endpoint of the EMILIA study was time to symptom worsening (time from randomization to the first documentation of a 5-point
`decrease from baseline) as measured by the Trial Outcome Index Physical/Functional/Breast (TOI-PFB) subset of the Functional
`Assessment of Cancer Therapy–Breast questionnaire. Predefined exploratory patient-reported outcome endpoints included propor-
`tion of patients with a clinically significant improvement in symptoms (per TOI-PFB) and proportion of patients with diarrhea symp-
`toms (per Diarrhea Assessment Scale). RESULTS: In the T-DM1 arm, 450 of 495 patients had a baseline and 1 postbaseline TOI-PFB
`score versus 445 of 496 patients in the capecitabine-plus-lapatinib arm. Time to symptom worsening was delayed in the T-DM1 arm
`versus the capecitabine-plus-lapatinib arm (7.1 months versus 4.6 months, respectively; hazard ratio 5 0.796; P 5.0121). In the T-DM1
`arm, 55.3% of patients developed clinically significant improvement in symptoms from baseline versus 49.4% in the capecitabine-
`plus-lapatinib arm (P 5.0842). Although similar at baseline, the number of patients reporting diarrhea symptoms increased 1.5- to
`2-fold during treatment with capecitabine and lapatinib but remained near baseline levels in the T-DM1 arm. CONCLUSIONS: Together
`with the EMILIA primary data, these results support the concept that T-DM1 has greater efficacy and tolerability than capecitabine
`plus lapatinib, which may translate into improvements in health-related quality of life. Cancer 2014;120:642–51. VC 2013 American Can-
`cer Society.
`
`KEYWORDS: ado-trastuzumab emtansine, T-DM1, HER2-positive, quality of life, breast cancer, antibody–drug conjugate.
`
`INTRODUCTION
`Trastuzumab emtansine (T-DM1) is an antibody–drug conjugate (ADC) comprising the humanized monoclonal antibody
`trastuzumab, a unique stable linker, and the highly potent cytotoxic agent DM1. In preclinical studies, it has been shown
`that T-DM1 delivers DM1 directly to human epidermal growth factor receptor 2 (HER2)–expressing tumor cells, where it
`is released intracellularly and causes apoptosis.1 Similar to trastuzumab, T-DM1 inhibits cell signaling through the phos-
`phatidylinositol 3-kinase/AKT pathway, inhibits HER2 shedding, and induces antibody-dependent cellular cytotoxicity.2,3
`Results were recently reported from the randomized phase 3 EMILIA study (TDM4370g/BO21977; Clinical-
`Trials.gov identifier NCT00829166) comparing T-DM1 with capecitabine plus lapatinib for patients with HER2-
`positive, locally advanced (LABC) or metastatic breast cancer (MBC) previously treated with a taxane and trastuzumab.
`Patients who were treated with T-DM1 had significantly longer progression-free survival (PFS; 9.6 months versus 6.4
`months; hazard ratio [HR] 5 0.65; P < .001) and overall survival (OS; 30.9 months versus 25.1 months; HR 5 0.68;
`
`Corresponding author: Manfred Welslau, MD, Medical Office Hematology, Am Hasenkopf 1,63739 Aschaffenburg, Germany, Bavaria; Fax: (011) 49 (0) 6021 3427-
`20; m.welslau@onkologie-ab.de
`
`1Medical Office Hematology, Aschaffenburg, Germany; 2Institut Curie, Paris, France; 3Yonsei University College of Medicine, Seoul, South Korea; 4UCLA Jonsson
`Comprehensive Cancer Center and Translational Research in Oncology, Los Angeles, California; 5Genentech, Inc., South San Francisco, California; 6Mount Vernon
`Cancer Centre, Northwood, United Kingdom
`
`*L. Fang is currently at Gilead Sciences, Inc, Foster City, California
`
`DOI: 10.1002/cncr.28465, Received: June 5, 2013; Revised: September 17, 2013; Accepted: October 1, 2013, Published online November 12, 2013 in Wiley
`Online Library (wileyonlinelibrary.com)
`
`642
`
`Cancer
`
`March 1, 2014
`
`IMMUNOGEN 2189, pg. 1
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`PROs From Phase 3 T-DM1 HER21 MBC Study/Welslau et al
`
`Figure 1. The EMILIA study design and PRO assessment schedule is illustrated. FACT-B assessments were performed on day 1 of
`cycle 1 and every second cycle thereafter until 6 weeks after PD and DAS assessments were performed on day 1 of each cycle dur-
`ing the treatment phase. Abbreviations: bid, twice a day; DAS, Diarrhea Assessment Scale; FACT-B, Functional Assessment of Can-
`cer Therapy–Breast; HER21, human epidermal growth factor receptor 2–positive; IV, intravenous; LABC, locally advanced breast
`cancer; MBC, metastatic breast cancer; PD, disease progression; PO, orally; PRO, patient-reported outcomes; q3w, once every 3
`weeks; qd, once a day; T-DM1, trastuzumab emtansine; tx, treatment. aFemale patients only. bApproximately every 3 months starting
`from the study drug completion visit until death, loss to follow-up, withdrawal of consent, or study discontinuation.
`
`P < .001), as well as fewer adverse events of grade 3 or
`greater, compared with those treated with capecitabine
`and
`lapatinib:
`41% (95% confidence
`interval
`[CI] 5 37%-45%) versus 57% (95% CI 5 53%-61%).4
`Based on these results, the US Food and Drug Adminis-
`tration recently approved T-DM1 for the treatment of
`HER2-positive MBC previously treated with trastuzumab
`and taxane chemotherapy.
`Despite the demonstrated activity and favorable
`safety profile associated with T-DM1, it is important to
`ensure that improved clinical outcomes do not come at
`the expense of a patient’s symptom burden. Patient-
`reported outcomes (PROs) provide a means of evaluating
`the subjective effect treatments have on patients and their
`quality of life (QOL).5 It is also important to understand
`the profile of adverse events from the patient’s perspective.
`Basch et al demonstrated that the incidence and severity
`of adverse events, as reported by investigators in clinical
`trials, can differ from reports of adverse events provided
`by patients.6,7 This difference can be particularly evident
`in cases of subjective adverse events, such as fatigue.
`PRO data have been reported from 2 previous phase
`2 studies of single-agent T-DM1 administered once every
`3 weeks (q3w). In TDM4374g, an open-label study of
`T-DM1 in patients with pretreated HER2-positive MBC
`(median of 7 prior agents for metastatic disease), PRO
`data suggested that T-DM1 is well tolerated and that
`treatment with T-DM1 does not increase symptom bur-
`den.8 In addition,
`in the randomized phase 2 study
`
`TDM4450g, the favorable safety profile associated with
`T-DM1 appeared to translate into superior overall QOL
`versus first-line treatment with trastuzumab and doce-
`taxel.9 To obtain a better understanding of PROs with
`T-DM1, a secondary endpoint of the pivotal phase 3
`EMILIA study was patient-reported time to symptom
`worsening. Here, we report the results of an EMILIA
`QOL analysis to evaluate the impact of T-DM1 and cape-
`citabine plus lapatinib on PROs.
`
`MATERIALS AND METHODS
`Study Design
`In the EMILIA study, patients with HER2-positive, unre-
`sectable LABC or MBC, who were previously treated with
`trastuzumab and a taxane, were randomized in a 1:1 ratio
`to T-DM1 (3.6 mg/kg intravenously q3w) or capecitabine
`(1000 mg/m2 orally [po] twice a day, on days 1 through
`14 of a q3w cycle) plus lapatinib (1250 mg po daily) until
`disease progression (PD) or unmanageable toxicity
`(Fig. 1). The study design and patient characteristics have
`been reported.4 Patients provided written informed con-
`sent; the study was approved by the relevant institutional
`review board and/or independent ethics committee at
`each study site.
`
`PRO Assessments
`of Cancer Therapy–Breast
`Functional Assessment
`(FACT-B) is a 37-item questionnaire with a 7-day recall
`period composed of 5 subscales that measure physical,
`
`Cancer
`
`March 1, 2014
`
`643
`
`IMMUNOGEN 2189, pg. 2
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`Original Article
`
`functional, social, and emotional well-being, as well as
`breast cancer–related symptoms. Female patients were
`asked to complete the FACT-B on day 1 of cycle 1 and
`every second cycle thereafter until 6 weeks after PD (or at
`the time of tumor assessment for patients who discontin-
`ued study treatment for reasons other than PD). The
`FACT-B was also given at the study drug completion visit
`and every 3 months thereafter at survival follow-up visits
`(Fig. 1).
`The FACT-B Trial Outcome Index Physical/Func-
`tional/Breast
`(TOI-PFB)
`is a 23-item subset of
`the
`FACT-B that can be used as a summary measure of physi-
`cal and functional well-being in patients with breast can-
`cer.10 Scores range from 0 to 92 points; a higher score
`indicates a better QOL. A change of 5 points is consid-
`ered clinically meaningful (ie, patients would have experi-
`enced clinically significant
`changes
`in health-related
`well-being).10-12 On this basis, the primary PRO endpoint
`in EMILIA was the time to symptom worsening as meas-
`ured by the FACT-B TOI-PFB, which was defined as the
`time from randomization to the first documentation of a
`decrease of 5 points from baseline.
`Predefined exploratory PRO endpoints in EMILIA
`included 1) the proportion of patients with a clinically sig-
`nificant improvement in symptoms between the 2 treat-
`ment arms as measured by the FACT-B TOI-PFB and 2)
`the proportion of patients with diarrhea symptoms as
`measured by the Diarrhea Assessment Scale (DAS). A
`patient was considered to have a clinically significant
`improvement in TOI-PFB score if the patient had at least
`one 5-point postbaseline increase in TOI-PFB score com-
`pared with baseline. The DAS is a 4-item questionnaire
`designed to evaluate the frequency, urgency, consistency,
`and discomfort of diarrhea experienced by a patient on a
`4-point scale13,14 with a 7-day recall period. The DAS was
`administered on day 1 of each cycle before the final analy-
`sis of PFS, after which the DAS was no longer adminis-
`tered. The incidence of each of the 4 diarrhea symptoms
`was calculated for both treatment arms at baseline and for
`each cycle in female patients with a baseline and 1
`follow-up assessment.
`Ad hoc exploratory (hypothesis-generating) PRO
`analyses included an evaluation of mean changes from
`baseline between treatment arms on the FACT-B physical
`well-being (PWB) subscale. The PWB subscale of the
`FACT-B contains 7 questions (scale 0 to 4) that measure
`physical symptoms related to adverse events. A difference
`of 5% of the scale range, or 0.25, is considered clinically
`meaningful, based on findings from published studies of
`minimally important differences in cancer scales.12,15
`
`Statistical Analyses
`For the primary PRO analysis, time to symptom worsening
`was assessed by Kaplan-Meier methods and a Cox model in
`and 1 postbaseline
`female patients with baseline
`TOI-PFB score, stratified by world region (United States,
`Western Europe, other); number of prior chemotherapeutic
`regimens for unresectable, locally advanced, or metastatic
`disease (0-1 versus > 1); and visceral versus nonvisceral
`disease. Two sensitivity analyses were performed for the
`primary analysis. In the first sensitivity analysis, symptom
`worsening that occurred after 1 or more missing assessments
`was backdated to the last nonmissing TOI-PFB assessment
`date plus 1 to assess the effect of missing assessments on the
`results. In the second sensitivity analysis, the date of symp-
`tom worsening was backdated by 6 weeks to assess the
`impact of the potential bias due to delayed reporting of
`symptom worsening. Log-rank P values were calculated,
`and the 95% CI for the median was computed using the
`method of Brookmeyer and Crowley.16
`Descriptive statistics (mean, standard error, mini-
`mum, and maximum) were used to evaluate the change
`from baseline in FACT-B TOI-PFB scores for each visit
`by treatment arm. In addition, a mixed-effects repeated
`measure model17 was used to evaluate the change from
`baseline in FACT-B TOI-PFB scores. Change from base-
`line in TOI-PFB score was the response variable; treat-
`ment, visit, and treatment-by-visit interaction terms were
`the fixed effects; baseline TOI-PFB score was the covari-
`ate; and within-subject correlation was modeled with a
`compound symmetry covariance structure. The treatment
`effects at week 12, 24, 36, and study drug termination vis-
`its were compared between the 2 treatment arms with the
`least-squares t tests in the mixed-effects model.
`The proportion of patients with a clinically signifi-
`cant improvement in TOI-PFB score in each treatment
`arm was estimated with 95% CIs calculated by using the
`Blyth-Still-Casella method.18,19 Treatment arms were
`compared, and a P value was computed using the strati-
`fied Mantel-Haenszel chi-squared test.20 For the DAS,
`summary statistics
`(frequency and proportion) by
`collapsed categories were used. Differences in the propor-
`tion of patients with moderate to severe diarrhea between
`treatment arms and 95% CIs were calculated.
`In an ad hoc exploratory analysis, a repeated measure
`mixed-effects model17 was used to estimate least squares
`means of change from baseline PWB sub-item scores.
`Treatment, time, and treatment-by-time interaction were
`fixed effects; baseline score was a covariate; and within-
`subject correlation was modeled with a compound
`symmetry covariance structure.
`
`644
`
`Cancer
`
`March 1, 2014
`
`IMMUNOGEN 2189, pg. 3
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`The compliance rates of TOI-PFB reporting were
`summarized by each visit. The compliance rates were calcu-
`lated as the ratio of the number of patients with nonmissing
`TOI scores divided by the number of patients expected to
`complete the FACT-B questionnaire as per the protocol.
`
`RESULTS
`Patient Characteristics
`randomized to the
`In total, 991 patients were
`capecitabine-plus-lapatinib (n 5 496) and the T-DM1
`(n 5 495) arms, respectively. Baseline demographics and
`disease characteristics were balanced between treatment
`arms.4 A total of 895 of 986 female patients had a baseline
`and 1 postbaseline TOI-PFB score, comprising 445 of
`496 patients in the capecitabine-plus-lapatinib arm and
`450 of 495 patients in the T-DM1 arm.
`
`Compliance With Completion of TOI-PFB
`Questionnaire
`The completion rates of TOI-PFB were 93.8% and
`89.3% at week 6 in the T-DM1 and capecitabine-plus-
`lapatinib arms, respectively. The completion rates for the
`T-DM1 arm were consistently higher than those reported
`in the capecitabine-plus-lapatinib arm at all time points
`except week 30. Thereafter, completion rates gradually
`declined to 71.4% and 70.0%, respectively, at week 48.
`Completion rates at the study drug completion visit were
`80.2% in the T-DM1 arm and 70.9% in the
`capecitabine-plus-lapatinib arm. Completion rates were
`lower during the survival follow-up period (after study
`drug discontinuation) and declined gradually at each
`3-month period: 50.3% and 48.0% for T-DM1 and cape-
`citabine plus lapatinib, respectively, at month 3, to 22.6%
`and 12.9%, respectively, at month 12 (Table 1).
`
`Primary Analysis: Time to Symptom Worsening
`Of the 450 patients in the T-DM1 arm and 445 patients in
`the capecitabine-plus-lapatinib arm who were eligible for
`the analysis of time to symptom worsening, 246 patients
`(54.7%) and 257 patients (57.8%), respectively, had wor-
`sening of symptoms. Median time to symptom worsening,
`as measured by the FACT-B TOI-PFB, was longer in
`the T-DM1 arm compared with the capecitabine-plus-
`lapatinib arm (7.1 months versus 4.6 months; HR 5 0.796,
`95% CI 5 0.667-0.951; P 5 .0121, based on the stratified
`analysis) (Fig. 2).4 The results of an unstratified analysis
`were consistent with these findings (data not shown).
`In addition, the results of 2 sensitivity analyses also
`demonstrated an increase in the median time to symp-
`tom worsening for patients who received T-DM1 com-
`
`PROs From Phase 3 T-DM1 HER21 MBC Study/Welslau et al
`
`TABLE 1. Compliance With Completion of the
`FACT-B TOI-PFB Questionnaire From Randomized
`Female Patients With Baseline and 1 Postbaseline
`Valid Score
`
`Patients With Nonmissing TOI-PFB Score
`Out of Patients Expected per Protocol
`
`Capecitabine 1
`Lapatinib
`(n 5 445), n/N (%)
`
`Trastuzumab
`Emtansine
`(n 5 450), n/N (%)
`
`Before Disease Progression
`Baseline
`445/445 (100)
`Week 6
`375/420 (89.3)
`Week 12
`318/377 (84.4)
`Week 18
`221/304 (72.7)
`Week 24
`158/216 (73.1)
`Week 30
`99/170 (58.2)
`Week 36
`96/138 (69.6)
`Week 42
`67/110 (60.9)
`Week 48
`56/80 (70.0)
`Study Drug
`210/296 (70.9)
`Completion Visit
`Survival Follow-up Period
`Month 3
`Month 6
`Month 9
`Month 12
`
`96/200 (48.0)
`51/158 (32.3)
`31/129 (24.0)
`15/116 (12.9)
`
`450/450 (100)
`405/432 (93.8)
`346/388 (89.2)
`289/344 (84.0)
`225/278 (80.9)
`116/232 (50.0)
`142/186 (76.3)
`98/161 (60.9)
`90/126 (71.4)
`203/253 (80.2)
`
`88/175 (50.3)
`54/134 (40.3)
`35/105 (33.3)
`19/84 (22.6)
`
`Abbreviations: FACT-B, Functional Assessment of Cancer Therapy–Breast;
`TOI-PFB, Trial Outcome Index Physical/Functional/Breast.
`
`Figure 2. Time to symptom worsening, based on randomized
`female patients with baseline and 1 postbaseline valid score.
`Abbreviations: Cap, capecitabine; CI, confidence interval; HR,
`hazard ratio; Lap, lapatinib; T-DM1, trastuzumab emtansine.
`
`pared with patients who received capecitabine plus
`lapatinib. The sensitivity analysis used to assess the
`impact of missing assessments resulted in a median time
`to symptom worsening of 6.0 months for the T-DM1
`arm versus 4.3 months
`for
`the capecitabine-plus-
`lapatinib arm (HR 5 0.788, 95% CI 5 0.660-0.941;
`P 5 .0089; stratified analysis). The analysis evaluating
`
`Cancer
`
`March 1, 2014
`
`645
`
`IMMUNOGEN 2189, pg. 4
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`Original Article
`
`Figure 3. Mean changes in FACT-B TOI-PFB scores of T-DM1 versus capecitabine plus lapatinib. Error bars indicate 95% confi-
`dence intervals. Abbreviations: FACT-B, Functional Assessment of Cancer Therapy–Breast; T-DM1, trastuzumab emtansine; TOI-
`PFB, Trial Outcome Index Physical/Functional/Breast.
`
`the impact of potential bias caused by delayed reporting
`of symptom worsening resulted in 6.6 months versus
`(HR 5 0.820,
`4.2 months,
`respectively
`95%
`CI 5 0.686-0.979; P 5 .0286; stratified analysis). These
`results are consistent with and support those observed in
`the primary analysis.
`
`Predefined Exploratory Analyses
`A descriptive summary of mean changes in FACT-B
`TOI-PFB scores from baseline per visit is shown in Figure
`3. In both arms, the average postbaseline TOI-PFB scores
`were slightly higher than baseline scores for visits before
`study drug termination but lower for visits after study
`drug termination. Patients in the T-DM1 arm reported
`more favorable TOI-PFB score changes compared with
`those in the capecitabine-plus-lapatinib arm for all but
`2 visits (week 48 on study drug and month 12 after study
`drug termination). The repeated measure mixed-effects
`model analysis showed a mean difference in TOI-PFB of
`1.30 points (95% CI 5 20.15 to 2.74; P 5 .0783) at
`(95% CI 5 20.83 to 2.55;
`week 12; 0.86 points
`P 5 .3195) at week 24; 1.28 points (95% CI 5 20.67 to
`3.15; P 5 .2023) at week 36; and 1.07 points (95%
`CI 5 20.59 to 2.73; P 5 .2068) at the drug termination
`visit, which all numerically favored T-DM1. In the
`
`mixed-effects model, the main effect of time and interac-
`tion effect of time by treatment were both statistically sig-
`nificant (P < .0001 and P 5 .0302, respectively).
`In the T-DM1 arm, 55.3% of patients (249 of 450;
`95% CI 5 50.7%-60.0%) had a clinically significant
`improvement in symptoms from baseline compared with
`49.4% of patients in the capecitabine-plus-lapatinib arm
`(220 of 445; 95% CI 5 44.7%-54.2%; P 5 .0842, based
`on a stratified analysis).
`Although similar at baseline, the number of patients
`reporting diarrhea symptoms increased 1.5- to 2-fold dur-
`ing treatment in the capecitabine-plus-lapatinib arm,
`whereas
`the number
`reporting diarrhea
`symptoms
`remained near baseline levels in the T-DM1 arm (Fig. 4).
`At cycle 2, more patients
`in the capecitabine-plus-
`lapatinib arm versus the T-DM1 arm had more than 1
`stool per day (60.5% versus 31.7%), had unformed stools
`(75.1% versus 39.5%), developed urgency (60.2% versus
`27.5%), and had abdominal discomfort (51.0% versus
`27.3%).
`
`Ad Hoc Exploratory Analysis
`In the FACT-B PWB subscale, there was a clinically
`meaningful improvement in the item “bothered by side
`effects” for patients in the T-DM1 arm compared with
`
`646
`
`Cancer
`
`March 1, 2014
`
`IMMUNOGEN 2189, pg. 5
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`PROs From Phase 3 T-DM1 HER21 MBC Study/Welslau et al
`
`Figure 4. Diarrhea Assessment Scale results are shown. Error bars indicate 95% confidence intervals. Abbreviation: T-DM1,
`trastuzumab emtansine.
`
`patients in the capecitabine-plus-lapatinib arm through
`week 24 (Fig. 5). T-DM1 also showed some benefit in all
`of the other items, although the differences were neither
`statistically nor clinically significant (Fig. 6).
`
`DISCUSSION
`Treatment of patients with MBC is often declared to be
`palliative in intent. Although studies of novel agents aim
`to show improvement in clinical outcomes, such as PFS
`and OS, rarely are improvements in symptoms or delays
`in symptom deterioration recorded in the context of phase
`3 clinical trials. Initially reported PRO results from phase
`2 studies8,9 are now validated in the phase 3 EMILIA trial.
`Here, we show that the safety and efficacy benefit of T-
`DM1 over standard-of-care combination chemotherapy
`and HER2-directed therapy, which itself has manageable
`toxicity, is accompanied by superior PROs. T-DM1 treat-
`ment resulted in a statistically significant delay in clini-
`cally meaningful symptom worsening (as measured by the
`first 5-point decrease from baseline in the TOI-PFB)
`compared with capecitabine plus lapatinib (7.1 months
`versus 4.6 months; HR 5 0.796; P 5 .0121).
`
`PROs are becoming increasingly important for eval-
`uating novel therapeutics in MBC, yet most studies have
`failed to demonstrate an appreciable difference between
`standard-of-care therapy and experimental agents.21-23
`Treatment with T-DM1 in EMILIA showed a statistically
`significant delay in symptom worsening and more favor-
`able mean changes from baseline in TOI-PFB scores for
`almost all visits compared with treatment with capecita-
`bine plus lapatinib. These differences were also reported
`when T-DM1 was compared with a different standard
`chemotherapy and HER2-directed combination, doce-
`taxel and trastuzumab, in the randomized phase 2 study
`TDM4450g.9 In the TDM4450g study, the time to
`a 5-point decrease in TOI-PFB score was significantly
`delayed in the T-DM1 arm, from a median of 3.5 months
`in the trastuzumab-plus-docetaxel arm to 7.5 months in
`the T-DM1 arm (HR 5 0.58, 95% CI 5 0.36-0.92;
`P 5 .022), and the mean changes from baseline in TOI-
`PFB scores were more favorable in the T-DM1 arm across
`all treatment cycles.9 Improved QOL, illustrated by the
`superior PRO results for patients treated with T-DM1
`shown in these 2 studies, is assumed to be because of the
`
`Cancer
`
`March 1, 2014
`
`647
`
`IMMUNOGEN 2189, pg. 6
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`Original Article
`
`Figure 5. Forest plot shows mean physical well-being scores of T-DM1 versus capecitabine plus lapatinib. Scores have been
`reverse-scaled (ie, higher score indicates better functioning). Abbreviations: Cap, capecitabine; CI, confidence interval; Lap, lapa-
`tinib; T-DM1, trastuzumab emtansine.
`
`unique dual mechanism of action of T-DM1 as an ADC;
`namely, the direct delivery of the cytotoxic agent DM1 to
`HER2-overexpressing cells combined with the specific
`antitumor activity of the trastuzumab portion of T-DM1.
`This innovative treatment approach may reduce systemic
`cytotoxic agent exposure and adverse events while also
`improving efficacy, especially compared with standard-of-
`care chemotherapy.
`In the EMILIA study, results from predefined ex-
`ploratory analyses also showed T-DM1 to be more favor-
`able than capecitabine plus lapatinib in terms of its impact
`on some health outcomes, including fewer diarrhea symp-
`
`toms (as assessed by the DAS). There was a trend favoring
`T-DM1 compared with capecitabine and lapatinib with
`regard to a clinically significant improvement in symp-
`toms (55.3% versus 49.4%; P 5 .0842). In addition, the
`results of the ad hoc exploratory analysis found that
`patients treated with T-DM1 were less bothered by side
`effects (as assessed by the FACT-B PWB subscale) com-
`pared with those who received capecitabine plus lapatinib.
`These results may be due, in part, to the fact that adverse
`events associated with T-DM1 were primarily laboratory
`abnormalities not associated with symptom deterioration,
`such as
`thrombocytopenia or elevated liver enzyme
`
`648
`
`Cancer
`
`March 1, 2014
`
`IMMUNOGEN 2189, pg. 7
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`PROs From Phase 3 T-DM1 HER21 MBC Study/Welslau et al
`
`Figure 6. Mean score by treatment arm on individual items comprising the physical well-being subscale of the FACT-B scores
`have been reverse-scaled (ie, a score of 4 indicates the best functioning and a score of 0 indicates the worst functioning). Error
`bars indicate 95% confidence intervals. Abbreviations: FACT-B, Functional Assessment of Cancer Therapy–Breast; T-DM1, trastu-
`zumab emtansine.
`
`Cancer
`
`March 1, 2014
`
`649
`
`IMMUNOGEN 2189, pg. 8
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`Original Article
`
`concentrations. In contrast, adverse events associated with
`capecitabine plus lapatinib were more likely to have physi-
`cal manifestations, such as diarrhea and palmar–plantar
`erythrodysesthesia.4
`Strengths of this PRO analysis include the fact that
`time to symptom worsening was a predefined secondary
`endpoint of the EMILIA study and several exploratory
`PRO analyses were defined a priori. Compliance with the
`TOI-PFB questionnaire was moderately high across all
`scheduled time points, and sensitivity analyses evaluating
`the impact of missed or delayed assessments support the
`main findings. Potential limitations in the design of the
`PRO analyses should be considered when interpreting the
`results, such as the open-label nature of the study, which
`could have led to reporting bias. In addition, missing data
`handling is a well-known challenge in PRO analysis.24
`The assumption that the missing data were missing at ran-
`dom in the PRO analysis may be violated, and bias could
`occur as a result. The data analysis in this article was con-
`ducted under the assumption that the missing data were
`ignorable (ie, missing at random or completely random)
`and should be interpreted with caution when nonignora-
`ble missing data are present. In reality, it cannot be ruled
`out that nonignorable missing data may exist. As dis-
`cussed by Siddiqui et al,25 the mixed-model approach is
`robust as long as the proportion of nonignorable missing
`data is reasonably small. The relatively high compliance
`rates observed in this study support the use of the mixed-
`model approach. Finally, in this study, PRO reporting
`occurred on day 1 of each cycle, and the results may have
`varied if a different reporting timeline was used. The dos-
`ing schedule and pharmacokinetics of the study drugs are
`not the same, and therefore symptoms may be experi-
`enced at different time points in the treatment cycle,
`which may have affected the results.
`Complementary to the results of the PRO analyses
`reported here are the results from other QOL-related anal-
`yses, which were designed to determine which treatment-
`related functional domains or symptoms matter most to
`patients or which adverse event patients most want to
`avoid. For example, a stated-preference (SP) instrument
`incorporating all functional domains and symptoms of
`the European Organization for Research and Treatment
`of Cancer QOL Questionnaire was administered to 150
`patients with breast cancer who had previously experi-
`enced chemotherapy.26 The SP survey asked patients to
`choose between a series of hypothetical functional/symp-
`tom pairs defined by combinations of health-related QOL
`characteristics. The 4 treatment effects that these patients
`most wished to avoid were nausea, vomiting, pain, and
`
`decreases in emotional and role functioning. Interestingly,
`in an online survey in which 551 respondents with MBC
`answered questions related to treatment decisions, the
`number of regimens patients received did not relate to
`their preference for future treatment.27 Furthermore,
`when choosing hypothetical treatments in the survey, in
`addition to treatment benefit, patients preferred treat-
`ments that did not cause alopecia. The information gath-
`ered from these SP questionnaires may facilitate
`discussion and encourage consideration of patient prefer-
`ences in decisions about treatment selection and could be
`used by clinical trial investigators to design more precise
`interventions to improve health-related QOL in those
`domains considered to be most important by patients.
`Together with the EMILIA primary results, these
`PRO data support the concept that the ADC T-DM1 has
`greater efficacy and tolerability than capecitabine plus
`lapatinib, which translates into improvements in health-
`related QOL.
`
`FUNDING SOURCES
`This study was sponsored by Genentech, Inc. Support for third-
`party writing assistance for this manuscript was provided by Genen-
`tech, Inc.
`
`CONFLICT OF INTEREST DISCLOSURE
`Drs. Welslau, Miles, and Dieras have served on advisory boards for
`Roche/Genentech; Drs. Miles and Dieras have participated in
`invited lectures for Roche/Genentech; Dr. Hurvitz has received
`research funding for her institution from Genentech and Glaxo-
`SmithKline, and travel reimbursement from Roche; Drs. Althaus
`and Guardino are full-time employees of Genentech and own
`Roche stock. Drs. Lalla and Fang were full-time employees of Gen-
`entech for the duration of the presented work and during the prepa-
`ration of the manuscript. Dr. Sohn made no disclosure.
`
`REFERENCES
`1. Lewis Phillips GD, Li G, Dugger DL, et al. Targeting HER2-
`positive breast cancer with trastuzumab-DM1, an antibody-cytotoxic
`drug conjugate. Cancer Res. 2008;68:9280-9290.
`2. Junttila TT, Li G, Parsons K, Phillips GL, Sliwkowski MX. Trastu-
`zumab-DM1 (T-DM1) retains all the mechanisms of action of tras-
`tuzumab and efficiently inhibits growth of
`lapatinib insensitive
`breast cancer. Breast Cancer Res Treat. 2011;128:347-356.
`3. Barok M, Tanner M, K€oninki K, Isola J. Trastuzumab-DM1 is
`highly effective in preclinical models of HER2-positive gastric can-
`cer. Cancer Lett. 2011;306:171-179.
`4. Verma S, Miles D, Gianni L, et al. Trastuzumab emtansine for
`HER2-positive advanced breast cancer. N Engl J Med. 2012;367:
`1783-1791. Erratum in N Engl J Med. 2013;368:2442.
`5. Lipscomb J, Gotay CC, Snyder CF. Patient-reported outcomes in
`cancer: a review of recent research and policy initiatives. CA Cancer
`J Clin. 2007;57:278-300.
`6. Basch E, Iasonos A, McDonough T, et al. Patient versus clinician
`symptom reporting using the National Cancer Institute Common
`Terminology Criteria for Adverse Events: results of a questionnaire-
`based study. Lancet Oncol. 2006;7:903-909.
`
`650
`
`Cancer
`
`March 1, 2014
`
`IMMUNOGEN 2189, pg. 9
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`PROs From Phase 3 T-DM1 HER21 MBC Study/Welslau et al
`
`7. Basch E, Jia X, Heller G, et al. Adverse symptom event reporting by
`patients vs clinicians: relationships with clinical outcomes. J Natl
`Cancer Inst. 2009;101:1624-1632.
`8. Agresta S, Hurvitz S, Lalla D, et al. Patient-reported outcomes
`(PROs) in HER21 metastatic breast cancer (MBC) treated with
`T-DM1. Ann Oncol. 2010;21(suppl 8):v111345. Abstract 1117.
`9. Hurvitz S, Dirix L, Kocsis J, et al. Phase II randomized study of
`trastuzumab emtansine versus trastuzumab plus docetaxel in patients
`with human epidermal growth factor receptor 2-positive metastatic
`breast cancer. J Clin Oncol. 2013;31:1157-1163. Erratum in J Clin
`Oncol. 2013;31:2977.
`10. Brady MJ, Cella DF, Mo F, et al. Reliability and validity of the
`Functional Assessment of Cancer Therapy-Breast quality-of-life
`instrument. J Clin Oncol. 1997;15:974-986.
`11. Yost KJ, Sorensen MV, Hahn EA, Glendenning GA, Gnanasakthy
`A, Cella D. Using multiple anchor- and distribution-based estimates
`to evaluate clinically meaningful change on the Functional Assess-
`ment of Cancer Therapy-Biologic Response Modifiers (FACT-BRM)
`instrument. Value Health. 2