Amgen Response to ICER’s Draft Evidence Report and Voting Questions on CAR-T Therapy for B-Cell Cancers

SUMMARY OVERVIEW

Amgen appreciates the opportunity to comment on ICER’s draft evidence report and voting questions on Chimeric Antigen Receptor T-Cell (CAR-T) for B-Cell Cancers. ICER compares tisagenlecleucel (Kymriah^TM [CTL019]) therapy to clofarabine-based therapy and blinatumomab-based therapy for patients up to 25 years of age with B-cell precursor acute lymphoblastic leukemia (B-ALL) that is refractory or in second or later relapse. This is an intensely critical point in patient treatment: having failed the initial line of therapy, children with relapsed/refractory (R/R) B-ALL face a median overall survival of 3 months.¹ Patient heterogeneity is extensive and its relationship to response is poorly understood.² On top of this, the therapeutic armamentarium of a handful of drugs is too small to address this huge patient and disease variation.³ Moreover, small numbers of patients in this orphan indication mean in the first 2 years after FDA approval, very little is known about new treatments (e.g., type of patients the treatment is best for, mechanism of action). It is at this point in tisagenlecleucel’s introduction that clinicians need real world experience and far more data to assess how and in which patients this treatment will save lives. The absence of adequate tisagenlecleucel clinical data negates the budgetary and cost-effectiveness evaluations which hinge on these data.

After a careful review of the draft report and consulting the opinions from practicing oncologists and hematologists, we remain concerned that this assessment is premature. This assessment is based on promising but inconclusive evidence and methodological flaws which may lead to results that are harmful to patients. The difficulty and imprecision in capturing value when there are too few patients alive or progression free is a commonly cited shortfall of value frameworks when applied to oncology.^4,5 Hence, ICER should delay this assessment until more conclusive evidence is available. In addition, ICER’s consideration of blinatumomab and tisagenlecleucel as mutually exclusive therapies is misaligned with real world clinical practice as these drugs are likely to be given sequentially.

B-ALL is an FDA-defined orphan disease with approximately 620 pediatric and young adult patients who relapse each year in the United States after achieving an initial response.⁶  This is well under the 10,000-prevalence threshold identified by ICER for its ultra-orphan framework. Assessing tisagenlecleucel as a non-orphan treatment fails to capture the complexities and distinctions faced in orphan drug development and use. Moreover, ICER’s decision to assess emicizumab for Hemophilia A under their ultra-orphan framework⁷ sets a strong precedent for ICER to conduct this R/R B-ALL assessment under this same ultra-orphan framework.

Our main recommendations on the draft report are summarized below:

1. The evidence base on tisagenlecleucel at this time is promising but inconclusive: ICER should rate the net health benefit as “I” (insufficient evidence) instead of “B+”, and wait for more data to perform a more accurate assessment.

2. Blinatumomab is not an appropriate comparator for tisagenlecleucel therapy in the pediatric R/R B-ALL population: ICER should remove it from the clinical effectiveness question.

3. The clinical effectiveness assessment in the leukemia setting lacks validity as it does not include all relevant studies:  ICER should ensure all relevant clinical studies are included in the analysis.

4. The patient heterogeneity between trials is not addressed: ICER should apply well-established methods to assess the comparative effectiveness of treatments.

5. The economic value assessment has methodological flaws, greatly underestimating the uncertainty of results: ICER should reanalyze survival outcomes and cost estimations and perform sensitivity analyses around the survival outcomes.

DETAILED COMMENTS AND RECOMMENDATIONS

1) The evidence base on tisagenlecleucel at this time is promising but inconclusive: ICER should rate the net health benefit as “I” (insufficient evidence) instead of “B+”, and wait for more data to perform a more accurate assessment.

ICER’s draft report points out the limitations in the current evidence base in 5 places: 1)“there are currently no randomized or observational trials directly comparing tisagenlecleucel therapy to salvage chemotherapy, any comparisons were at substantial risk for selection bias” (p. 10). 2)“All of the clinical studies were small (less than 100), single arm designs with limited follow up” (median follow up less than one year) and incomplete reporting” (p.19); 3) “most of the pivotal trials of CAR-Ts have yet to be published in peer reviewed journals.”(p. 25); 4) “Both the benefits and duration of and long-term relapse-free survival is unknown at this point (p. 40)”; and 5)“uncertainty around long-term harms of therapy,…which makes the analyses versus standard therapy controversial”(p. 40). These important flaws and limitations in the evidence base introduce a low certainty in the net health benefit of tisagenlecleucel. This invalidates the draft report’s conclusion: “there is at least a small net health benefit compared with current salvage chemotherapy although the benefit may be substantial (“B+” rating)”(p.35).

According to ICER’s framework, an insufficient (I) rating should be granted in any situation in which the level of certainty in the evidence is low. Strictly adhering to ICER’s value framework,⁸ tisagenlecleucel should be rated as insufficient evidence (I) rating. Given this rating, per the United States Preventive Services Task Force guidelines, this makes the answer ‘No’ to ICER’s question, “Is the evidence adequate to demonstrate a net health benefit for treatment with tisagenlecleucel (Kymriah™, Novartis) versus treatment with clofarabine or comparable chemotherapy (e.g., blinatumomab, multi-agent chemotherapy including clofarabine)?”.  Assuming the panel also recognizes the insufficient evidence and votes ‘No’, then by default, it negates any subsequent votes on the value of tisagenlecleucel, given that the evidence base has been deemed promising but too inconclusive to determine value. The development of questions and the subsequent panel voting process must be designed and navigated carefully to ensure an accurate capture of value in order to best serve patients, their families and providers.

In addition, blinatumomab is not a “chemotherapy” and has demonstrated significant survival benefit over salvage chemotherapy in adult patients with R/R B-ALL in a randomized, controlled study.⁹ Blinatumomab is a first-in-class, bispecific T-cell engager (BiTE^®) antibody construct that binds specifically to CD19 expressed on the surface of cells of B-lineage origin, and to CD3 expressed on the surface of T cells. We ask ICER to correct the terminology throughout the evidence report and the voting questions.

2) Blinatumomab is not an appropriate comparator for tisagenlecleucel therapy in the pediatric R/R B-ALL population: ICER should remove it from the clinical effectiveness question (first voting question for the pediatric B-ALL population, page 1).

Based on feedback from oncologists and hematologists who treat these patients every day, in the real-world setting it is highly likely that tisagenlecleucel therapy and blinatumomab will be used sequentially to produce the deepest remission possible.^10,11 Moreover, blinatumomab has been used in patients who failed tisagenlecleucel on trial, and it is likely to be used this way in clinical practice.  The cornerstone of B-ALL treatment is combination therapy rather than a discrete choice between treatments. Several oncology and hematology practitioners, who are also opinion leaders, have confirmed the sequential use of blinatumomab and tisagenlecleucel therapy in their practice. These insights suggest that ICER’s draft report comparing these treatments is likely irrelevant to real world clinical practice and assesses a scenario that may never exist.

3) The clinical effectiveness assessment in the leukemia setting lacks validity as it does not include all relevant studies: ICER should ensure all relevant clinical studies are included in the analysis.

If ICER proceeds to conduct the assessment with blinatumomab as a comparator, ICER should include all relevant clinical studies (see Appendix A). Amgen agrees with ICER on the need to rely heavily on grey literature given the evolving evidence base. However, the studies included in the draft report for the comparators are not comprehensive. According to ICER’s policy on inclusion of grey literature in evidence reviews, grey literature includes conference proceedings and/or abstracts. As mentioned in the two data packages sent by Amgen to ICER, for blinatumomab studies in the pediatric R/R B-ALL population, in addition to the already included MT103-205 study, another expanded access study of blinatumomab in the pediatric R/R B-ALL population (RIALTO, NCT02187354) should have been included. The RIALTO study data sent to ICER was presented at the ASCO 2017 annual conference.¹² RIALTO has many similarities to MT103-205, providing a significant increase in the number of subjects, with almost the same length of follow up as MT103-205 of blinatumomab for pediatric patients with R/R B-ALL. It is worth noting that the baseline inclusion criteria on blast level of RIALTO is identical to that of the ELIANA study, i.e., ≥ 5%, whereas MT103-205 required the baseline blast level to be >25%. Therefore, pooling the RIALTO study with MT103-205 data will help to increase the number of patients that share similar characteristics to tisagenlecleucel patients and allow for more precise estimates. Endpoints of RIALTO are consistent with MT103-205 as well, including: complete remission (CR) within the first two cycles; relapse-free survival (RFS); overall survival (OS) and rate of hematopoietic stem cell transplantation (HSCT) after CR.

In assessing the efficacy of tisagenlecleucel, the draft report includes three single-arm studies of tisagenlecleucel in the pediatric R/R B-ALL population (B2101J with split dosing of tisagenlecleucel; standard dose study B2205J; and standard dose pivotal study B2202/ELIANA). At the end of the clinical effectiveness assessment, tisagenlecleucel data are pooled across all three trials to estimate the long-term survival of patients. We disagree with this approach as the B2101J study had a completely different dosing regimen than the other two studies. This severely biases ICER’s tisagenlecleucel efficacy estimation.

In contrast to presenting pooled data on three tisagenlecleucel studies in assessing efficacy, only study B2202/ELIANA data is evaluated to assess the harms. This approach is both incomplete and misleading in that the B2101J study was primarily designed to assess the safety, tolerability and engraftment potential of tisagenlecleucel in pediatric R/R B-ALL.

4) The patient heterogeneity between trials is not addressed: ICER should apply well-established methods to assess the comparative effectiveness of treatments.

As stated in ICER’s draft report, there are no randomized or observational trials directly comparing tisagenlecleucel therapy to the comparators, making any comparisons subject to major risk for selection bias. The draft report attempts to describe the study sample characteristics, but this fails to address the selection bias problem. Instead, ICER should apply existing methods that have been used extensively in assessing treatment comparative effectiveness.^13,14,15

The ELIANA¹⁶ and MT-103-205¹⁷ studies in pediatric R/R B-ALL patients are all small single-arm studies enrolling quite heterogeneous patients:

• Age differences: ELIANA enrolled older pediatric patients up to 3 to 21 years of age, while the MT103-205 enrolled younger patients from 0 to 18 years old.
• Baseline bone marrow blast level differences: ELIANA required baseline blast levels at enrollment to be greater than or equal to 5%, whereas MT103-205 required baseline blast levels to be greater than 25%.
• Previous treatment history differences: In ELIANA, the median number of previous lines of therapy was 3.0, meaning that at least 50% of patients had 3 or more prior lines of therapy; whereas in MT103-205, 49 patients (70%) had only 1 or 2 prior therapies. ELIANA contained only 21% of patients with refractory disease compared to 56% of patients in MT103-205. ELIANA required patients to have life expectancy longer than 12 weeks, but MT103-205 did not have any such requirements.

In addition, the patient characteristics described in the ICER report do not include all the important potential factors that might determine the outcomes. Using the method applied in Quinn et al.,^18,19 we found that patients in MT103-205 and RIALTO on average had a 29% higher risk of mortality than an average patient in the ELIANA study. This method considered available baseline characteristics from both blinatumomab and tisagenlecleucel trials. Among the prognostic factors explored (age and age squared, gender, race, primary refractory, chemotherapy refractory, prior HSCT, previous lines of therapies), many of these proved to be impactful factors on the risk of mortality. For example, refractory status is an important factor that impacts OS. Of note, 52 (34.2%) patients in MT103-205 and RIALTO were chemotherapy refractory,²⁰ whereas 9 (10%) of ELIANA study enrolled patients were chemotherapy refractory (Appendix A). These results highlight the heterogeneity of patient characteristics between blinatumomab and tisagenlecleucel studies and the critical importance of adjusting for that heterogeneity in comparing these studies.

Given the aforementioned differences in patient characteristics, the indirect treatment comparison needs to adjust for all effect modifiers and prognostic factors: for reference on methods for population-adjusted indirect comparisons, please see NICE’s recent technical support document.²¹ Without this analysis, the conclusions on treatment effect and causation may be a function of other unrelated and coincidental variables. 

5) The economic value assessment has methodological flaws, greatly underestimating the uncertainty of results: ICER should reanalyze survival outcomes and cost estimations and perform sensitivity analyses around the survival outcomes.

Survival outcomes are the most important model inputs in cost-effectiveness (CE) models in oncology. Therefore, the estimations of OS/event-free survival (EFS) and the sensitivity of the model around OS/EFS should be examined carefully.  In the current ICER model, the OS/EFS results of tisagenlecleucel are derived by pooling the three single-arm studies, which is inappropriate; as mentioned above, study B2101J should not be included.

In addition, it appears that the ICER model assumes no further relapse after 13 months and no further B-ALL related death after 30 months, an assumption that is not based on any evidence and further inconsistent with the 4-year cure assumption mentioned in the method section of the report. The current CE model sensitivity analyses (one-way and probabilistic) does not include sensitivity around survival curve parameters. Failure to incorporate uncertainty in the survival distribution estimates renders the sensitivity analyses results virtually meaningless, as the survival distributions represent the main source of uncertainty in the model. Moreover, survival curves for immuno-oncologic agents differ from chemotherapy, cautions need to be taken while selecting the parametric distributions for survival.^22,23 Therefore, the current conclusion that the CE model results are robust through one-way and probabilistic sensitivity analyses is unfounded.

Two other cost assumptions are also biased in favor of tisagenlecleucel. In the base case, ICER assumes that the cost of tisagenlecleucel is only applicable for responders at one month. Although this might be appropriate for publicly insured patients per the manufacturer’s public statement, it is unclear if the payment strategy will be the same for privately insured patients. Also, the markup rate applied in the model is not evidence-based given the high degree of uncertainty both due to marked differences between hospitals and the confidential nature of real mark-up rates. In addition, the cost per day for hospital stay is estimated based on HCUP estimation for all children, which is an underestimation for the pediatric R/R B-ALL patients. The model also underestimates the cost of B-cell aplasia.

ADDITIONAL COMMENTS/CORRECTIONS

In addition, we have identified a total of 80 areas of concerns and/or errors that we ask ICER to address in their Revised Report; 13 of these are methodological concerns that need to be addressed and 67 are factual errors or inaccuracies that need to be corrected.  Please see Appendix B for details, including suggested corrections.

CONCLUSION

The need for continued innovation in R/R B-ALL is reflected in the low number of available treatments. Tisagenlecleucel is a welcomed option desperately important to this extremely vulnerable pediatric patient population. A fundamental tenant in the application of value frameworks is in their timing, and despite ICER taking on this assessment to raise this important dialogue, the timing is premature and the assessment should be reconsidered. If ICER decides to proceed with this assessment, then the draft report needs to address the above noted limitations and concerns, including acknowledging insufficient evidence (I) for tisagenlecleucel and removing blinatumomab as a comparator to better align with real world clinical practice.

Appendix A: Patient characteristics in tisagenlecleucel and blinatumomab clinical studies

Appendix B: Other Comments and Corrections

Based on our detailed review, we identified the following methodological flaws or concerns that we ask ICER to address as they incorporate changes for their Revised Report. 

1. On page 7, the report states “blinatumomab, which has been used as a bridge to SCT with some success”. This is not an accurate statement in that blinatumomab has been shown to improve patients’ outcomes independent of SCT²⁴ and should not be presumed to be limited to SCT eligible patients as a bridging therapy.

2. On page 11, Table 1.1, Key outcomes and harms, the rationale for selecting this list of outcomes and harms is not provided. The anxiety over the receipt of a novel treatment based on genetically modified cells should be considered as harms. On page 13, “some of the patients will die and others will become too sick to tolerate treatment with the CAR-T cells” should be included as a potential harm of treatment.

3. On page 11, the evidence on intervention effectiveness and harms is derived from studies with a median duration of at least three months. No rationale is provided for this requirement.

4. On page 16, Section 3.2, Clinical guidelines, fails to mention blinatumomab is recommended by NCCN guidelines as a category 1 designated therapy and the fact that blinatumomab is recommended by NICE within its marketing authorization as an option for treating Philadelphia-chromosome-negative relapsed or refractory precursor B-cell acute lymphoblastic leukemia in adults.²⁵

5. On page 25, “it was not possible to estimate the comparative benefits or harms of these novel therapies…using either direct or indirect comparisons”. This statement is incorrect. It is possible to conduct unanchored indirect comparisons using single-arm studies. As such, on page 27, “since none of the studies included comparator groups, we were unable to perform any statistical comparisons…”  The absence of a comparator group does not eliminate the possibility of statistical comparisons.

6. On page 26, the study selection fails to describe how the comparator studies were identified.

7. On page 28, Table 4.1, Summary of treatments for relapsed/refractory pediatric B-ALL, this table presents patient characteristics based on the infused patients. Given that patients assigned to tisagenlecleucel could die/experience manufacture failures and/or AEs waiting for tisagenlecleucel therapy and not all patients were infused, it is more appropriate to compare the characteristics of the enrolled patients.  ICER’s report underpins this on page 24 with, “The reported overall remission rates for tisagenlecleucel in the three trials (from 69% to 95%, Table 4.2) represents an optimistic presentation of the results that violates the intention to treat principle because they are based on patients who received successful infusion of CAR-T cells, thereby excluding patients who did not receive the therapy because of manufacturing failures, death prior to infusion, or AEs.”

8. On page 29, “thus, patient selection suggested that the patients in the trials of tisagenlecleucel had undergone more prior therapies and, thus, had a worse prognosis at enrollment”. This statement is unfounded and not evidence based. The precise direction and relative importance of the factors for predicting survival in these patients has not been established.  This statement fails to account for other potential differences in patient characteristics between trials including patient age and percent of patients who are relapsed, which were higher in the tisagenlecleucel than blinatumomab trials and which, according to analyses of data from blinatumomab trials, are both favorable prognostic factors.

9. On page 31, Table 4.4, Estimated event-free survival at six months in therapies for relapsed or refractory childhood B-ALL, the estimates of EFS at 6 months for tisagenlecleucel reported in Table 4.4 appear to be calculated by multiplying reported estimates of EFS at 6 months among infused patients from Table 1-2 of the Novartis FDA Briefing Document by the ratio of the number of infused vs. enrolled patients in Study 2101J, 2505J, and 2202. This approach yields EFS estimates for these studies of 58%, 46%, and 53% respectively, which are the same as the estimates reported by ICER.

EFS was not evaluated in the 205 trial of blinatumomab. The value reported (16%) appears to be the product of the % with CR (39%) from Table 4.2 from the ICER report and the estimated RFS at 6 months (42%) reported by Von Stackelberg et al. (2016). RFS was not reported for 2101J, 2205J, or 2202 so it is not feasible to replicate this precise calculation for these studies. However, the Kaplan Meier estimate of duration of remission (DOR) may be a reasonable proxy for RFS for these studies.  Accordingly, the value for Study 2202 obtained by multiplying the % CR among enrolled patients with >=3 months FU reported by ICER (63%) in Table 4.2 by the Kaplan Meier estimate of DOR at 6 months reported in Table 1-2 the Novartis FDA Briefing Document (75.4%) is 47%. This value, which is arguably more comparable to the 16% reported for blinatumomab, is significantly lower than the 53% reported for Study 2202 by ICER. The corresponding calculated values for Studies 2101J and 2205J are 54% and 38%, respectively, which also are lower than the values of 58% and 46% reported by ICER. 

10.  On page 30, the report says the CR rate for B2101J study enrolled cohort is 52/71=73% (61%-83%) and does not specify the source. This number is incorrect. It is reported in the ODAC briefing document that the number of patients achieving ORR=52; CR=38 (p. 51). Therefore, the CR rate for the enrolled cohort should be 38/71=54% and ORR rate for the enrolled cohort should be 52/71=73%.

11.  On page 30, the report says the CR for B2205J study enrolled cohort is 20/35=57% (39%-74%) and does not specify the source. This number is incorrect. It is reported in the ODAC briefing document that the number of patients achieving ORR is 20, CR is 18 (p. 51). Therefore, the CR rate for the enrolled cohort should be 18/35=51%; the ORR for the enrolled cohort should be 20/35=57%.

12.  On page 30, the report says that the CR rate for the B2202 study enrolled cohort should be 52/83=63% (51%-73%) and doesn't specify the source. The number is incorrect. It is reported in the FDA briefing document for the ODAC 2017 Tisagenlecleucel meeting that the number of patients achieving ORR=52, CR=40 (p. 39). In B2202 study, 88 patients are enrolled, it is not 83 (5 patients should NOT be excluded from the enrolled cohort due to lack of follow-up). Therefore, the CR rate for the enrolled cohort should be 40/88=45%; the ORR rate for the enrolled cohort should be 52/88=59%.

13.  On page 31, the report states that the ORR in the enrolled population in the B2202 study is 65.8% (95% CI 54%-76%). It is unclear how 65.8% is obtained. There are 52 patients who achieved ORR in B2202. Among all the enrolled patients, the ORR should be 52/88=59.1%; among enrolled patients and patients with at least 3 month follow up, the ORR should be 52/83=63%. 

In addition, we identified these factual errors/inaccuracies in the draft report. We ask that ICER make corresponding corrections and reflect them in the Revised Report.

14.  On page 14, definition of the complete remission, neutrophils>1×109/L should be corrected to 1×10⁹/L, same for platelets.

15.  On page 96, Table C1, the report says the median follow-up duration for B2202 study is 4.8 months. According to the ODAC briefing document, the median follow-up duration of response = 4.8 months. Median follow-up EFS = 5.6 months. It is unclear why 4.8 months is selected over 5.6 months. Source: Oncologic Drugs Advisory Committee Briefing Document. Tisagenlecleucel (CTL019) for the Treatment of Pediatric and Young Adult Patients with Relapsed/Refractory B-Cell Acute Lymphoblastic Leukemia. 2017

16.  On page 96, Table C1,the report says the prior lines of chemo for 2205J is 3 and references the Buechner 2017 publication. This is the wrong reference. The number appears in FDA Briefing Document Oncologic Drugs Advisory Committee Meeting. BLA 125646 Tisagenlecleucel Novartis Pharmaceuticals Corporation.

17.  On page 96, Table C1, the report says the number of patients enrolled in B2101J is 71 and references the Maude 2015 and Grupp 2013 publications. The references are incorrect. The numbers are presented in the Oncologic Drugs Advisory Committee Briefing Document. Tisagenlecleucel (CTL019) for the Treatment of Pediatric and Young Adult Patients with Relapsed/Refractory B-Cell Acute Lymphoblastic Leukemia. 2017.

18.  On page 96, Table C1, the report says the number of patients infused in the B2101J study is 55 and references the Maude 2015 paper. The source is incorrect. The number appears in Oncologic Drugs Advisory Committee Briefing Document. Tisagenlecleucel (CTL019) for the Treatment of Pediatric and Young Adult Patients with Relapsed/Refractory B-Cell Acute Lymphoblastic Leukemia. 2017.

19.  On page 96, Table C1, the report says the median follow-up for the B2101J study is 7 months for the 30 patients and references Maude 2015 and Grupp 2013 publications. The sources are incorrect. The correct source should be Maude SL, Frey N, Shaw PA, et al. Chimeric antigen receptor T cells for sustained remissions in leukemia. The New England journal of medicine. 2014;371(16):1507-1517.

20.  On page 96, Table C1, the report says the age in B2101J study is 11 years and references Maude 2015 and Grupp 2013 publications. The sources are incorrect. The number is presented in Oncologic Drugs Advisory Committee Briefing Document. Tisagenlecleucel (CTL019) for the Treatment of Pediatric and Young Adult Patients with Relapsed/Refractory B-Cell Acute Lymphoblastic Leukemia. 2017.

21.  On page 96, Table C1, the report says number of prior lines of chemotherapy for B2101J is 4 and references Maude 2015 and Grupp 2013 publications. The sources are incorrect. The number is presented in Oncologic Drugs Advisory Committee Briefing Document. Tisagenlecleucel (CTL019) for the Treatment of Pediatric and Young Adult Patients with Relapsed/Refractory B-Cell Acute Lymphoblastic Leukemia. 2017.

22.  On page 96, Table C1, the report says the proportion of patients with prior SCT in the 2101J study (n=30) cohort is 72%. The number is incorrect. The 72% is actually among the pediatrics cohort (n=25) as reported in the Maude et al., 2014 NEJM. The correct number should be 63.6% as reported in the Oncologic Drugs Advisory Committee Briefing Document. Tisagenlecleucel (CTL019) for the Treatment of Pediatric and Young Adult Patients with Relapsed/Refractory B-Cell Acute Lymphoblastic Leukemia. 2017.

23.  On page 96, Table C1, the report says the median follow-up for the MT103-205 study of blinatumomab is >2 years. However, the correct number should be 23.8 months (Von Stackelberg et al., 2016, JCO).

24.  On page 29, the report says “There is no accepted definition of a cure, as relapses can rarely occur more than 10 years after remission” and references the Pui et al., 2003 publication. This is an incorrect interpretation of the publication, where it says “Our results suggest a new working definition of cure: 10 or more years of continuous complete remission, a standard that could be used to gauge the effectiveness of current and future treatment plans”.

25.  On page 30, Table 4.2, the report says the ORR for blinatumomab in the MT103-205 study is 45% and the CR rate is 39%. This is incorrect. The 39% reported in the von Stakelberg 2016 publication is the ORR rate.

26.  On page 31, the report states that the B2202 study requires patients to have >5% blasts in bone marrow at screening. It should be ≥5%.

27.  On page 33, the report says that the incidence of CRS (all grades) in B2202 study (n=68) is 79% and references the Novartis ODAC 2017 document. The correct source should be the Kymriah^TM FDA package insert, 2017 (p. 7).

28.  On page 33, the report says that the incidence of CRS, grade ≥3, in the B2202 study (n=68) is 49% and references the Novartis ODAC 2017 document. The correct source should be the Kymriah^TM FDA package insert, 2017 (p. 7).

29.  On page 33, the report says the incidence of neurologic toxicities (all grades) in B2202 study (n=68) is 65%, same comment as above.

30.  On page 33, the report says the incidence of grades ≥3 neurologic toxicities in B2202 study (n=68) is 18%, same comment as above.

31.  On page 33, the report says the incidence of all grades encephalopathy in B2202 study (n=68) is 34%. Same comment as above.

32.  On page 33, the report says the incidence of grades ≥3 encephalopathy in B2202 study (n=68) is 10%, same comment as above.

33.  On page 33, the report says the incidence of all grades headache in B2202 study (n=68) is 37%, same comment as above.

34.  On page 33, the report says the incidences of grades ≥3 headache in B2202 study (n=68) is 3%, same comment as above.

35.  On page 33, the report says the incidence of all grades acute kidney injury in B2202 study (n=68) is 22%, same comment as above.

36.  On page 33, the report says the incidence of grades ≥3 acute kidney injury in B2202 study (n=68) is 13%, same comment as above.

37.  On page 33, the report says the incidence of grades ≥3 hypotension in B2202 study (n=68) is 22%, same comment as above.

38.  On page 33, the report says the incidence of grades ≥3 hypoxia in B2202 study (n=68) is 18%, same comment as above.

39.  On page 33, the report says the incidence of all grades infections with unknown pathogens in B2202 study (n=68) is 41%, same comment as above.

40.  On page 33, the report says the incidence of grades ≥3 infections with unknown pathogens in B2202 study (n=68) is 16%, same comment as above.

41.  On page 33, the report says the incidence of all grades viral infections in B2202 study (n=68) is 26%, same comment as above.

42.  On page 33, the report says the incidence of grades ≥3 viral in B2202 study (n=68) is 18%, same comment as above.

43.  On page 33, the report says the incidence of all grades bacterial infections in B2202 study (n=68) is 19%, same comment as above.

44.  On page 33, the report says the incidence of grades ≥3 bacterial in B2202 study (n=68) is 13%, same comment as above.

45.  On page 33, the report says the incidence of all grades fungal infections in B2202 study (n=68) is 13%, same comment as above.

46.  On page 33, the report says the incidence of grades ≥3 fungal infections in B2202 study (n=68) is 7%, same comment as above.

47.  On page 33, the report says the incidence of grades ≥3 disseminated intravascular coagulation in B2202 study (n=68) is 9%, same comment as above.

48.  On page 33, the report says the incidence of grades ≥3 histiolymphocytic hemophagocytosis in B2202 study (n=68) is 7%, same comment as above.

49.  On page 33, the report says the incidence of grades ≥3 heart failure in B2202 study (n=68) is 7%, same comment as above.

50.  On page 33, the report says the incidence of grades ≥3 cardiac arrest in B2202 study (n=68) is 4%, same comment as above.

51.  On page 33, the report says the incidence of grades ≥3 seizures in B2202 study (n=68) is 3%, same comment as above.

52.  On page 33, the report says the incidence of grades ≥3 intracranial hemorrhage in B2202 study (n=68) is 1%, same comment as above.

53.  On page 33, the report says the incidence of all grades fever in B2202 study (n=68) is 50% and references the Novartis ODAC 2017 document. The number presented in the source document is 40% (p. 65).

54.  On page 33, the report says the incidence of grades ≥3 fever in B2202 study (n=68) is 15% and references the Novartis ODAC 2017 document. The source reports that incidence for serious fever is 7% (p. 66).

55.  On page 97, Table C2, the report states that the inclusion criteria for B2101J is “relapsed and refractory CD 19+ cancers B-ALL in 1^st to 4^th relapse; 3 refractory primary B-ALL” and references the Maude et al., 2015 Blood publication. This information is not presented in the Maude et al., 2015 publication. Another publication of the study (Maude et al., 2014 NEJM) mentions that “26 had B-cell ALL in the first to fourth relapse, 3 had primary refractory B-cell ALL”.

56.  On page 98, Table C2, the report says that the MT103-205 study of blinatumomab includes patients with ≥25% blasts in bone marrow, it should be >25%.  The report says the study includes patients refractory or in 1st subsequent relapse, this is incorrect. It should be “primary refractory, in 1st relapse after full salvage induction regimen, in second or later relapse, or in any relapse after alloHSCT”. The report says the study includes B-ALL, it should be B-precursor ALL.

57.  On page 99, the report says that the median weight (kg) in B2202 study is 43 and references the Kymriah^TM package insert. This number is not reported in the source.

58.  On page 99, the report says that the baseline performance status for the B2202 is 90 and references the Kymriah^TM package insert. The number is not reported in that source.

59.  On page 99, the report says that 12% of the B2202 patients are chemorefractory, 9% are primary refractory, 79% are relapse disease and references the Kymriah^TM package insert. The numbers are not reported in the source.

60.  On page 99, the report says that 85% of the B2202 patients receive bridging chemotherapy and references the Kymriah^TM package insert. The number is not reported in the source.

61.  On page 99, the report says that 45% of the patients in B2101J are female and references the Maude 2014 NEJM publication. The number reported is incorrect. In the Maude 2014 publication, it reports 44% of the pediatrics cohort are female, whereas 46% is reported in the ODAC 2017 briefing document.

62.  On page 99, the report says that “87% in 1^st-4^th relapse 60%” in study B2101J and references the Maude 2014 NEJM publication. The numbers reported in the publication are 100% with >=1 relapse and 0% primary refractory for the pediatrics cohort (p. 1509).

63.  On page 99, the report says that 64% of the B2101J patients have prior SCT and references the Maude 2014 NEJM publication. The number reported in the source is 72% in the pediatric cohort.

64.  On page 101, the report says the median OS for B2202 study is 16.6 months and references the Kymriah^TM package insert. The number is not reported in the source.

65.  On page 101, the report says the rate of PR is 7.0% (87.4-77.4) and references the Kymriah^TM package insert. The number is not reported in the source. In addition, the number does not make sense as the point estimation is not included in the confidence interval.

66.  On page 101, the report says the % dead before response assessment in the B2202 study is 7.5% and references the Kymriah^TM package insert. The number is not reported in the source.

67.  On page 101, the report says the % of non-responders in the B2202 study is 7.9% (n=8) for N=63 and references the Kymriah^TM package insert. The numbers are not reported in the source.

68.  On page 101, the report says the proportion of patients receiving allo-SCT in the B2202 study is 10.5% and references the Kymriah^TM package insert. The number is not reported in the source, where it states that the “stem cell transplantation rate among those who achieved CR/CRi is 12% (6/52).”

69.  On page 101, the report says the median OS for the B2101J study is 32.7 months and references the Maude 2014 NEJM publication. The number is not reported in the source.

70.  On page 101, the report says the % of CR in the B2101J study is 69% and references the Maude 2014 NEJM publication. Same comment as above.

71.  On page 101, the report says the % of non-responders in the B2101J study is 5.5% and references the Maude 2014 NEJM publication. Same comment as above.

72.  On page 101, the report says the % of ORR in the B2101J study is 95% and references the Maude 2014 NEJM publication. Same comment as above.

73.  On page 101, the report says the % of non-responders of blinatumomab is 55% and references the von Stackelberg JCO 2016 publication. The number should be 30% (p. 4).

74.  On page 102, the report says the proportion of patients with grade three/four AEs in B2202 study infused cohort is 65% and references the Kymriah^TM package insert. The number is not reported in the source.

75.  On page 102, the report says the proportion of patients with treatment-related death in the B2202 study infused cohort is 17% total death and references the Kymriah^TM package insert. The number is not reported in the source.

76.  On page 102, the report says the proportion of patients with prolonged B-cell aplasia in B2202 study infused cohort is 84% and references the Kymriah^TM packager insert. The number is not reported in the source.

77.  On page 102, the report says the proportion of patients with grade three/four CRS in the B2101J study is 17% and references the Maude 2014 NEJM publication. The number reported in the source is 27% (severe CRS, 8 patients, p. 1507).

78.  On page 102, the report says the proportion of patients with prolonged B-cell aplasia in the B2101J study is 90%  and references the Maude 2014 NEJM publication. The number is not reported in the source.

79.  On page 102, the report says the proportion of patients with neurotoxicity in the MT103-205 blinatumomab study is not reported and references the von Stackelberg JCO 2016 publication. The number should be 24% (17/70).

80.  On page 102, the report says the proportion of patients with grade three/four neurotoxicity in the MT103-205 blinatumomab study is not reported and references the von Stackelberg JCO 2016 publication. The number should be 4% (3/70) (p. 4).

REFERENCES