Boceprevir for the treatment of genotype 1 chronic hepatitis C

Transcription

1 MSD Hertford Road Hoddesdon Hertfordshire EN11 9BU UK Telephone +44 (0)1992 xxxxxx Facsimile +44 (0)1992 xxxxxxx Kate Moore Technology Appraisals Project Manager National Institute for Health and Clinical Excellence Level 1A, City Tower, Piccadilly Plaza Manchester M1 4BD 7 th September 2011 Dear Kate Boceprevir for the treatment of genotype 1 chronic hepatitis C Please find below the response to the Evidence Review Group (ERG) questions 1. We have also corrected additional errors as described later as well as provided a revised base case analysis. Kind regards xxxxxx xxxxxxxxxxxxx xxxxxxxxxxxxxxxxxxx 1 Black text = original questions. Blue text = MSD responses. Red text = corrected/additional values in tables

2 Section A: Clarification on effectiveness data A1. Please provide a list of references excluded at the full-text screening stage. The references excluded at full-text screening stage are provided below and copies of the papers are attached. Davis G, Krawczynski K, Szabo G. Hepatitis C virus infection pathobiology and implications for new therapeutic options. Dig Dis Sci. 2007;52: Guilloche P, Feray C. Systematic review: anti-viral therapy of recurrent hepatitis C after liver transplantation. Aliment Pharmacol Ther. 2011;33: A2. Please give p values for Tables B13, B15, B16 and B17. The p values for Tables B13, B15, B16 and B17 are provided in the amended tables below. P values were calculated using the Chi 2 test in all cases, except where one or more value was < 5, when the Fishers Exact test was used ('F' appended to value where this used) Table B 13: Achievement of SVR by Response at TW 4 and 8 in Treatment Experienced Patients (FAS) Group 1 (PR48) N = 80 RESPOND-2 Group 2 (BOC/PR RGT) N = 162 Group 3 (BOC/PR48) N = 161 SVR by TW 4 Response, n/n (%) (p value vs ) P05685 (PEG2a/RControl) N = 67 Group 2 (BOC/PEG2a/R) N = 134 Undetectable HCV RNA or decreased by 1log 10 IU/ml 17/67 (25.4) 80/110 (72.7) 90/114 (78.9) 14/57 (24.6) 79/112 (70.5) HCV RNA decreased by < 1log 10 IU/ml 0/12 15/46 (32.6) p=0.0251(f) 15/44 (34.1) p=0.0240(f) 0/9 7/18 (38.9) p=0.0593(f) SVR by TW 8 Response, n/n (%) (p value vs ) Undetectable HCV RNA 7/7 (100) 64/74 (86.5) p= /84 (88.1) p= /9 (44.4) 64/72 (88.9) p=0.0044(f) Detectable HCV RNA 8/65 (12.3) 29/72 (40.3) p= /70 (42.9) 9/56 (16.1) 21/50 (42.0) p=0.0031

3 Table B 15: End of Treatment Response Rates and Relapse Rates in Treatment Experienced Patients (FAS) RESPOND-2 P05685 (PR48) N = 80 Group 2 (BOC/PR RGT) N = 162 Group 3 (BOC/ PR48) N = (PEG2a/R) [Control] N = 67 Group 2 (BOC/PEG2a/R) N = 134 Undetectable HCV RNA at end of treatment, n (%) p value vs 25/80 (31.3) 114/162 (70.4) 124/161 (77.0) 28/67 (41.8) 99/134 (73.9) Relapse Rate, n (%) p value vs 8/25 (32.0) 17/111 (15.3) p= /121 (11.6) p= /21 (33.3) 11/95 (11.6) p=0.0127

4 Table B 16: Achievement of SVR in Treatment Naïve Patients With and Without Cirrhosis SPRINT-2 SPRINT-1 SVR in Patients Without Cirrhosis n/n (%) p value vs SVR in Patients With Cirrhosis n/n (%) p value vs SVR in Patients with Missing Data Regarding Cirrhosis n/n (%) p value vs Non-Black Cohort Black Cohort Combined Cohorts Entire Cohort Group 1 (PR48) N = /287 (39.7) 6/13 (46.2) 5/11 (45.5) Group 2 (BOC/PR RGT) N = /292 (69.2) 5/13 (38.5) p= /11 (36.4) p=1(f) Group 3 (BOC/ PR48) N = /279 (69.9) 9/22 (40.9) p= /10 (90.0) p= Group 1 (PR48) N = 52 12/52 (23.1) Group 2 (BOC/PR RGT) N = 52 20/45 (44.4) p= /0 0/3 p=na(f) 0/0 2/4 (50.0) p=na(f) Group 3 (BOC/ PR48) N = 55 28/52 (53.8) p= /2 (50.0) p=na(f) 0/1 p=na(f) Group 1 (PR48) N = /339 (37.2) 6/13 (46.2) 5/11 (45.5) Group 2 (BOC/PR RGT) N = /337 (65.9 5/16 (31.3) p= /15 (40.0) p= Group 3 (BOC/ PR48) N = /331 (67.4) 10/24 (41.7) p= /11 (81.8) p= (PR48) N = /96 (38.5) 2/8 (25.0) Group 2 (PR4/ PRB24) N = /96 (56.3) p= /7 (57.1) p=0.3147(f) 0/0 0/0 0/0 Group 3 (PR4/ PRB44) N = /97 (76.3) 3/6 (50.0) p=0.5804(f)

5 Table B 17: Achievement of SVR in Treatment Experienced Patients With and Without Cirrhosis RESPOND-2 P05685 (PR48) N = 80 Group 2 (BOC/PR RGT) N = 162 Group 3 (BOC/ PR48) N = (PEG2a/RControl) N = 67 Group 2 (BOC/PEG2a/R) N = 134 SVR in Patients Without Cirrhosis n/n (%) p value vs 16/66 (24.2) 85/132 (64.4) 85/128 (66.4) 11/53 (20.8) 70/101 (69.3) SVR in Patients With Cirrhosis n/n (%) p value vs 0/10 (0.0) 6/17 (35.3) p=0.0570(f) 17/22 (77.3) (F) 1/9 (11.1) 12/24 (50.0) p=0.0560(f) SVR in Patients with Missing Data Regarding Cirrhosis n/n (%) p value vs 1/4 (25.0) 4/13 (30.8) p=1(f) 5/11 (45.5) p=0.6044(f) 2/5 (40.0) 4/9 (44.4) p=1(f)

6 A3. Please clarify why N values for the 'n/n's for each outcome within the groups do not add up to the total N for each group in Tables B12 and B13, and also Table B17 Respond-2 Group 3. Tables B12 and B13 in the submission dossier did not include the number of patients with missing data, therefore the patient numbers shown in the table did not add up to the total N for each group. This information has been added to Table B12 (see below) and B13 (see question A2). Please note that Table B15 was mislabelled 'naïve' instead of 'experienced' and also that the total number of patients for Group 3 in the Table heading should be 161 not 311. These errors have been corrected in the version of Table B15 provided above in response to Question A2. A further check of the remaining tables in this section revealed that the same numerical error in the Table header regarding total number of patients in Group 3 had been repeated in Table B17 (also corrected above). Please note that these table header errors do not impact on any of the other values in the tables or calculations from derived from them.

7 Table B12: Achievement of SVR by Response at TW 4 and 8 in Treatment Naive Patients (FAS) SPRINT-2 Non-Black Cohort Black Cohort Combined Cohorts (PR48) N = 311 Group 2 (BOC/PR RGT) N = 316 Group 3 (BOC/ PR48) N = 311 (PR48) N = 52 SVR by TW 4 Response, n/n (%) (p value vs ) Undetectable HCV RNA or decreased by 1log 10 IU/ml 121/234 (51.7) 187/228 (82.0) p < /218 (81.7) p < /26 (46.2) Group 2 (BOC/PR RGT) N = 52 16/24 (66.7) p = 0.17 Group 3 (BOC/ PR48) N = 55 22/36 (61.1) p = 0.30 (PR48) N = /260 (51.2) Group 2 (BOC/PR RGT) N = /252 (80.6) p < Group 3 (BOC/ PR48) N = /254 (78.7) p < HCV RNA decreased by < 1log 10 IU/ml 3/62 (4.8) Missing 1/15 (6.7) 21/73 (28.8) p < /15 (20.0) 31/79 (39.2) p < /14 (28.6) SVR by TW 8 Response, n/n (%) (p value vs ) Undetectable HCV RNA Detectable HCV RNA 48/56 (85.7) 73/233 (31.3) Missing 4/22 (18.2) 170/190 (89.5) p = /104 (36.5) p = /22 (13.6) 166/182 (91.2) p = /102 (43.1) p = /27 (13.6) 0/21 6/24 (25.0) p = /16 (31.3) p = /83 (3.6) 0/5 0/4 2/3 (66.7) 1/20 (5.0) 3/4 (75.0) 8/38 (21.1) 1/10 (10.0) 14/18 (77.8) p = /25 (32.0) 18/22 (81.8) p = /29 (27.6) p = 0.38 p = /9 3/4 (75.0) 51/60 (85.0) 81/271 (29.9) 5/32 (15.6) 27/97 (27.8) p < /19 (15.8) 184/208 (88.5) p = /129 (35.7) p = /31 (9.7) 36/95 (37.9) p < /17 (35.3) 184/204 (90.2) p = /131 (39.7) p = /31 (19.4)

8 A4. Please provide clarification of the meta-analysis methods and results as no summary measure is given and the results seem to be simple averages by treatment group. In addition, on page 95 it is stated that no important heterogeneity between studies was found. Please clarify if this means that no statistically significant heterogeneity was found, and please provide the z and p values. Clarification of methods for calculating proportions provided by Professor Mills (the external analyst). "Statistically, binomial proportions are not simple. Several different approaches can be used to calculate the confidence intervals for a proportion 2. For the purpose of this meta-analysis, we used an approach we developed with the makers of StatsDirect, led by Professor Iain Buchan, that first transforms proportions into a quantity (the Freeman-Tukey variant of the arcsine square root transformed proportion, 3,4 y = arcsine* (r/(n +1)++arcsine * (r +1)/(n +1)+, with a variance of 1/(n +1), where n is the denominator for population size). This is suitable for the usual fixed and random effects meta-analysis. We have used this approach in several high-impact peer-reviewed manuscripts 5,6,7,8 and this approach has been widely accepted by many other meta-analysis groups in a variety of domains 9. Given the small number of studies (n=2), it is unsurprising that the chosen approach is similar to other possible approaches. We confirmed this by examining both simple averages and a Bayesian exact method." In the tables below a summary measure (relative risk) is provided for the results, both for the fixed effects and random effects model. Heterogeneity was assessed using z-tests with P-values, Q-values, and the I2. In all different approaches to measuring 2 Newcombe R. Two sided confidence intervals for the single proportion: a comparative evaluation of seven methods. Statistics in Medicine 1998;17: Freeman M, Tukey J. Transformations related to the angular and the square root. Annals of Mathematical Statistics. 1950;21: Jaynes ET. Confidence intervals and Bayesian intervals. In Harper WL, Hooker CA, eds. Foundations of Probability Theory, statistical Inference, and Statistical Theories of Science. Dordrecht, The Netherlands: Reidel; Mills EJ, Nachega JB, Buchan I, Orbinski J, Attaran A, Singh S, Rachlis B, Wu P, Cooper C, Thabane L, Wilson K, Guyatt GH, Bangsberg DR. Adherence to antiretroviral therapy in sub-saharan Africa and North America: a meta-analysis. JAMA. 2006;296: Mills EJ, Nachega JB, Bangsberg DR, Singh S, Rachlis B, Wu P, Wilson K, Buchan I, Gill CJ, Cooper C. Adherence to HAART: a systematic review of developed and developing nation patient-reported barriers and facilitators. PLoS Med. 2006;3:e438 7 Ford N, Mofenson L, Kranzer K, Medu L, Frigati L, Mills EJ, Calmy A. Safety of efavirenz in firsttrimester of pregnancy: a systematic review and meta-analysis of outcomes from observational cohorts. AIDS. 2010;24: Mills E, Eyawo O, Lockhart I, Kelly S, Wu P, Ebbert JO.Smoking cessation reduces postoperative complications: a systematic review and meta-analysis. Am J Med. 2011;124: e8. 9 Pal T, Permuth-Wey J, Kumar A, Sellers TA. Systematic review and meta-analysis of ovarian cancers: estimation of microsatellite-high frequency and characterization of mismatch repair deficient tumor histology. Clin Cancer Res. 2008;14:

9 Population heterogeneity, no significant heterogeneity between studies was found. This is supported by the near identical estimates derived from the fixed effects versus random effects analysis. Assessment of treatment effects when combining the RESPOND-2 trial with the P05685 trial - fixed effects analysis Relative Risk Overall SVR 0.32 ( ) Null 0.13 responders ( ) Cirrhotics 0.12 ( ) Relative Risk Reduction 68% (55-77%) 87% (5-98%) 88% (44-98%) Relative risk (RR) when greater than 1 is positive 3.10 ( ) 7.42 ( ) 8.55 ( ) Q-value I 2 z-score P-value for noncombin ability Absolute risk reduction % (47-74%) % % (40-79%) % % (23-60%) Population Assessment of treatment effects when combining the RESPOND-2 trial with the P05685 trial - random effects analysis Relative risk Overall SVR 0.32 ( ) Null 0.14 responders ( ) Cirrhotics 0.14 ( Relative Risk Reduction 68% (55-77%) 86% (6-98%) 86% (32-97%) Relative risk (RR) when greater than 1 is positive 3.10 ( ) 7.37 ( ) 6.91 ( ) Q-value I 2 z-score P-value for noncombina bility Absolute risk reduction % % (47-74%) % % (17-63%) % % (2-80%) A5. On page 103, in the adverse events section, it is stated that anaemia and dysgeusia were the only two events that were reported with a 10% difference in the BOC/PEG2b/R arms compared with the pooled PEG2b/R control arms of the three studies. Please clarify why only adverse events reported with a 10% difference between groups have been commented on as differing between groups were these the only statistically significant differences in adverse events reported between the two groups? All adverse events (AEs) which occurred in 10% of patients in the boceprevir pooled safety analysis are reported in Table B23 in the original document. In the text surrounding the table we have only commented on those adverse events which occurred with a 10% difference between treatment groups in the pooled analysis (anaemia and dysgeusia). This approach was taken so as to be consistent with that used in the regulatory submission dossier to the EMA. For completeness, information regarding the statistical significance of the difference in the incidence of AEs between the two groups is provided in response to question A6.

10 A6. Please provide the 95% confidence intervals for the relative risk of each adverse event in Tables B23, B24 and B25. Please also provide information on the risk difference for each adverse event in these tables. Tables B23, B24 and B25 have been updated to include 95% CI for the relative risk and the risk difference for each AE. Table B23: Adverse Events Reported in the Pooled Safety Analysis (Incidence 10%) PEG2b/R a (n=547) BOC/PEG2b/R (n=1548) RR (95% CI) RD (95% CI) n % n % Treatment-Emergent AE % % 1.01 (1.00, 1.02) 1.18 (-0.01, 2.37) Treatment-Related, % % 1.02 (1.00, 1.03) 1.53 (0.11, 2.94) Treatment-Emergent AE Serious AE % % 1.35 (0.98, 1.86) 2.73 (0.01, 5.46) Death b % % 0.35 (0.09, 1.41) (-1.23, 0.28) Life-Threatening % % 1.11 (0.48, 2.59) 0.14 (-0.97, 1.25) Study Drug % % 1.08 (0.84, 1.40) 0.99 (-2.23, 4.22) Discontinuation Due to AE Dose Modification Due to AE c % % 1.62 (1.38, 1.90) (10.62, 19.28) Blood And Lymphatic System Disorders Anaemia % % 1.69 (1.47, 1.94) (15.35, 24.43) Neutropenia % % 1.29 (1.05, 1.58) 5.06 (1.25, 8.87) Gastrointestinal Disorders Diarrhoea % % 1.25 (1.02, 1.52) 4.52 (0.67, 8.38) Dry Mouth % % 1.21 (0.90, 1.62) 1.92 (-0.98, 4.82) Dysgeusia % % 2.45 (1.98, 3.02) (17.87, 25.54) Nausea % % 1.12 (1.00, 1.26) 4.90 (0.11, 9.69) Vomiting % % 1.60 (1.23, 2.07) 6.54 (3.31, 9.77) General Disorders And Administration Site Conditions Asthenia % % 0.90 (0.73, 1.11) (-5.46, 1.91) Chills % % 1.13 (0.98, 1.31) 3.84 (-0.65, 8.32) Fatigue % % 1.01 (0.93, 1.10) 0.39 (-4.43, 5.22) Influenza Like Illness % % 0.89 (0.75, 1.06) (-6.94, 1.38) Injection Site Erythema % % 0.89 (0.68, 1.17) (-4.41, 1.86)

11 Injection Site % % 1.03 (0.77, 1.37) 0.30 (-2.69, 3.29) Reaction Irritability % % 1.00 (0.83, 1.20) (-4.07, 3.95) Pain % % 1.25 Pyrexia % % 1.02 (0.88, 1.18) 0.62 (-3.89, 5.12) Investigations Weight Decreased % % 0.97 (0.74, 1.27) (-3.43, 2.73) Metabolism And Nutrition Disorders Decreased Appetite % % 1.09 (0.91, 1.30) 2.08 (-2.04, 6.21) Musculoskeletal And Connective Tissue Disorders Arthralgia % % 1.11 (0.89, 1.37) 1.76 (-1.89, 5.42) Myalgia % % 0.97 (0.81, 1.16) (-4.84, 3.41) Nervous System Disorders Dizziness % % 1.27 (1.00, 1.61) 3.67 (0.22, 7.11) Headache % % 1.03 (0.92, 1.15) 1.34 (-3.49, 6.17) Psychiatric Disorders Anxiety % % 1.12 (0.85, 1.47) 1.30 (-1.78, 4.39) Depression % % 0.99 (0.81, 1.20) (-4.11, 3.64) Insomnia % % 1.04 (0.90, 1.20) 1.09 (-3.43, 5.61) Respiratory, Thoracic And Mediastinal Disorders Cough % % 0.91 (0.74, 1.12) (-5.41, 2.05) Dyspnoea % % 1.22 (0.98, 1.51) 3.40 (-0.23, 7.02) Dyspnoea Exertional % % 1.2 Skin And Subcutaneous Tissue Disorders Alopecia % % 1.03 (0.87, 1.21) 0.69 (-3.57, 4.94) Dry Skin % % 1.14 (0.92, 1.42) 2.26 (-1.37, 5.89) Pruritus % % 0.92 (0.77, 1.11) (-5.86, 2.27) Rash % % 0.97 (0.78, 1.20) (-4.17, 3.07) Note: Patients may have had more than one adverse event. a Excludes events for 36 patients in SPRINT-1 after they crossed over from Arm 1 (PR) to BOC/PR. b Deaths are also included in the Serious AE count c Excludes patients who discontinued due to adverse events

12 Table B24: Nadir Haematological Laboratory Values During the Treatment Phase, by modified WHO Category, in the Pooled Safety Analysis WHO Grade PEG2b/R a (n=547) BOC/PEG2b/R (n=1548) n % n % Number of Patients Included b Haemoglobin (g/dl) RR (95% CI) a Excludes events for 36 patients in SPRINT-1 after they crossed over from Arm 1 (PR) to BOC/PR. b Only subjects with at least one treatment value for a given laboratory test are included. RD (95% CI) % % 0.53 (0.47, 0.60) (-26.59, ) 9.5 to < % % 1.20 (1.06, 1.35) 7.37 (2.59, 12.15) 8.0 to < % % 1.87 (1.50, 2.33) (9.13, 16.60) 6.5 to < % % 2.06 (0.93, 4.57) 1.37 (0.13, 2.62) < % % (0.01, 0.51) 8.5 to < 10 NA % % 1.66 (1.42, 1.93) (12.29, 21.14) < 8.5 NA % % 2.20 (1.31, 3.69) 3.55 (1.66, 5.44) Number of Patients Included b WBC (10 9 /L) > % % 0.64 (0.54, 0.76) (-14.38, -5.87) 2.0 to % % 0.86 (0.77, 0.96) (-11.06, -1.37) 1.5 to < % % 1.42 (1.18, 1.70) 8.71 (4.60, 12.81) 1.0 to < % % 2.16 (1.50, 3.09) 6.84 (4.24, 9.43) < % % 3.17 (0.74, 13.61) 0.80 (0.06, 1.54) Number of Patients Included b Neutrophils (10 9 /L) > % % 0.59 (0.49, 0.71) (-14.17, -6.09) 1.0 to % % 0.87 (0.76, 1.00) (-8.95, 0.24) 0.75 to < % % 1.09 (0.92, 1.30) 2.19 (-2.03, 6.41) 0.5 to < % % 1.67 (1.32, 2.11) 8.88 (5.34, 12.41) < % % 1.84 (1.18, 2.87) 3.42 (1.29, 5.54) Number of Patients Included b Platelets (10 9 /L) > % % 0.79 (0.75, 0.83) (-22.16, ) 70 to % % 1.93 (1.49, 2.50) (6.85, 13.53) 50 to < % % 3.88 (2.11, 7.14) 5.90 (4.09, 7.70) 25 to < % % 3.29 (1.32, 8.23) 2.13 (0.95, 3.31) < % % (-0.03, 0.42)

13 Table B25: Treatment-Related, Treatment-Emergent Adverse Events Reported in P05685 (Incidence 10%) Treatment- Emergent AE Treatment-Related, Treatment- Emergent AE PEG2a/R a (n=67) BOC/PEG2a/R (n=134) n % n % RR (95% CI) % % RD (95% CI) % % 1.02 (0.99, 1.05) 1.49 (-1.41, 4.40) Serious AE % % 1.29 (0.56, 2.93) 2.99 (-6.34, 12.31) Death b % (-0.56, 3.55) Life-Threatening % (0.52, 6.94) Study Drug Discontinuation Due to AE % % 3.83 (1.19, 12.31) (4.61, 20.77) Dose Modification % % 1.93 (1.19, 3.14) (7.86, 33.93) Due to AE c Blood And Lymphatic System Disorders Anaemia % % 1.52 (1.04, 2.23) (3.09, 31.24) Leukaemia % % 5.00 (1.20, 20.76) (4.66, 19.22) Leukopenia Neutropenia % % 1.75 (0.99, 3.10) (1.35, 25.52) Gastrointestinal Disorders Diarrhoea % % 3.00 (1.22, 7.38) (5.47, 24.38) Dysgeusia % % 2.60 (1.41, 4.78) (12.01, 35.75) Nausea % % 1.42 (0.90, 2.22) (-2.23, 24.62) Vomiting % General Disorders And Administration Site Conditions Asthenia % % 1.17 (0.63, 2.15) 2.99 (-8.49, 14.46) Chills % % 0.88 (0.39, 1.98) (-10.83, 7.84) Fatigue % % 0.92 (0.69, 1.21) (-19.11, 10.16) Influenza Like Illness % % 0.94 (0.58, 1.54) (-14.41, 11.43) Irritability % % 0.91 (0.53, 1.55) (-14.60, 10.12) Pyrexia % % 1.21 (0.53, 2.78) 2.24 (-7.00, 11.48)

14 Metabolism And Nutrition Disorders Decreased Appetite % % 1.04 (0.56, 1.94) 0.75 (-10.56, 12.05) Musculoskeletal And Connective Tissue Disorders Arthralgia % % 0.64 (0.31, 1.32) (-16.24, 4.30) Muscle spasms % % Myalgia % % 2.50 (1.00, 6.24) (2.08, 20.31) Nervous System Disorders Dizziness % % 0.94 (0.44, 2.01) (-10.67, 9.17) Headache % % 0.83 (0.53, 1.31) (-18.59, 8.14) Anxiety % % 0.78 (0.35, 1.70) (-12.65, 6.68) Depression % % 1.58 (0.66, 3.78) 5.22 (-3.81, 14.26) Insomnia % % 0.73 (0.44, 1.18) (-21.20, 4.78) Sleep Disorder % % 2.50 (0.75, 8.34) 6.72 (-0.57, 14.00) Respiratory, Thoracic And Mediastinal Disorders Cough % % 0.85 (0.46, 1.57) (-14.34, 8.37) Dyspnoea % % 0.81 (0.47, 1.41) (-16.69, 7.73) Dyspnoea Exertional Skin And Subcutaneous Tissue Disorders % % Alopecia % % 2.20 (0.87, 5.55) 8.96 (0.07, 17.84) Dry Skin % % 0.91 (0.46, 1.79) (-12.22, 9.23) Pruritus % % 1.13 (0.52, 2.45) 1.49 (-8.18, 11.17) Rash % % 3.00 (1.22, 7.38) (5.47, 24.38) Note: Patients may have had more than one adverse event. a Excludes events for 36 patients in SPRINT-1 after they crossed over from Arm 1 (PR) to BOC/PR. b Deaths are also included in the Serious AE count c Excludes patients who discontinued due to adverse events

15 Section B: Clarification on cost-effectiveness data B1. Please provide the source for the distribution of patients per fibrosis level before treatment described in Table B31 (page 144). The distribution of the patients per fibrosis level before treatment as reported in Table B31 was estimated on the basis of the baseline characteristics of the cohorts studied in the clinical studies SPRINT-2 and RESPOND-2. as reported in the clinical study reports and reproduced below. Extract from Table 7 - Baseline characteristics SPRINT-2 (CSR p.140) Extract from Table 7 - Baseline characteristics RESPOND-2 (CSR p.116)

16 For the calculation of the proportion of patients included for each fibrosis level, the missing values were subtracted from the total number of the patients, producing the denominator for our calculations as reported below: Naïve Experienced Number of studied patients Number of patients for which data is missing Total number of patient included in the economic analysis The estimation of the distribution of the patients per fibrosis level as reported in Table B31 of the report is being presented below (patient numbers bolded). Distribution of patients per fibrosis level before treatment Naïve Experienced F_0 4.43% (47/1060) 4.80% (18/375) F_ % (730/1060) 53.33% (200/375) F_ % (183/1060) 21.07% (79/375) F_3 4.43% (47/1060) 7.73% (29/375) F_4 5.00% (53/1060) 13.07% (49/375) B2. On page 166 of the manufacturer s submission, it is stated that the METAVIR scoring system was applied in the economic analysis in order to maintain consistency with the methodology used in the clinical trials and the SPRINT-2 and the RESPOND-2. Please clarify which specific studies are meant here is it just SPRINT-2 and RESPOND-2 or are other clinical trials and the meta-analysis included? The phrase in the "clinical trials and the SPRINT-2 and the RESPOND-2 is a typographical error, it should read as "in the clinical trials SPRINT-2 and RESPOND-2. B3. Also in relation applying the METAVIR scoring system in the economic analysis (page 166), the following inconsistencies between the Summary of Product Characteristics for boceprevir and the clinical trials are not explained: The marketing authorisation has different treatment strategies and different stopping rules than those used in the clinical trials

17 The characteristics of the patient groups considered in the marketing authorisation differ from the trials participant groups in terms of their initial fibrosis level. That is, whereas all groups in the trials are reported to have a mix of fibrosis levels discriminated into two groups - F0-F2 and F3-F4 fibrosis levels, the marketing authorisation indicates the assignment of treatment regimens according to different initial fibrosis levels F0-F3 and cirrhotic patients (F4). Please provide the rationale for these discrepancies. [ Further clarification from ERG: Pages provide a description of the difference between the stopping rules and treatment strategies in the SPC and trials. However, the rationale for these differences is not clear (the why rather than the what ). Please could you provide an explanation in order to clarify if it is reasonable to assume that the same treatment effect can be expected in both situations.] The decision to recommend different treatment strategies, stopping rules, and to group patients differently according to their fibrosis level, compared to the clinical trials, was made by the EMA when granting marketing authorisation for boceprevir. The rationale for their decisions is provided below. All of the information below is taken from the EPAR, which can be accessed via the EMA website ( Treatment Strategies The CHMP assessed the response guided therapy strategies used in the phase III clinical trials and in some cases decided that sufficient uncertainty surrounded the shortened treatment durations, causing them to take the conservative approach of recommending longer treatment durations for some groups. In no cases did they recommend a shorter duration of therapy than that in the trial, therefore it is expected that boceprevir regimens administered as recommended in the SPC would have equivalent or greater efficacy compared to the regimens in the trials. Treatment Naïve Patients Early Responders The EPAR states that treatment naïve patients who received the shortened treatment duration RGT arm (28 weeks) in SPRINT-2 achieved high SVR rates which were similar to those observed with 48 weeks of treatment. Therefore the CHMP recommend that early responders receive 4 weeks of PEG/R lead-in followed by 24 weeks BOC/PEG/R, as in the SPRINT-2 clinical trial. Treatment Naïve Patients Late Responders Lower SVR rates were observed among late responders who received the long duration RGT arm in SPRINT-2 (4 weeks PEG/R, 24 weeks, BOC/PEG/R, 20 weeks PEG/R) compared to the matched population who received the BOC/PEG/R48 arm (4 weeks

18 PEG/R, 44 weeks BOC/PEG/R). This was largely due to higher viral breakthrough while patients were receiving PEG/R dual therapy for the last 20 weeks. The EPAR states that 24 weeks of boceprevir is too short for this group of patients, and uses the data for late responders from RESPOND-2 to estimate the optimal treatment duration. In RESPOND-2, treatment experienced late responders who received the long duration RGT arm (4 weeks PEG/R, 32 weeks BOC/PEG/R, 12 weeks BOC/PEG/R) achieved similar SVR and similar relapse rates to those who received BOC/PEG/R48 (4 weeks PEG/R, 44 weeks BOC/PEG/R). In addition, no excess of viral breakthroughs was observed when boceprevir was administered for 32 weeks, suggesting there is no benefit in extending boceprevir therapy beyond 32 weeks in these patients. The CHMP recommend 32 weeks of boceprevir as part of a 48 week treatment regimen for treatment naïve late responders. Justification for the applicability of this RESPOND- 2 data to the treatment naïve population is given on pages of the EPAR. Treatment Experienced Patients Early Responders In RESPOND-2, SVR rates were lower in early responders who received the short duration RGT arm (4 weeks PEG/R, 32 weeks BOC/PEG/R) compared to the matched population who received the BOC/PEG/R48 arm (4 weeks PEG/R, 44 weeks BOC/PEG/R), suggesting these patients should receive therapy for longer than 36 weeks. As mentioned previously, no value is expected from extending therapy beyond 32 weeks in treatment experienced late responders, and the CHMP infered that this applies to treatment experienced early responders as well. Therefore it is recommended that early responders who have received prior therapy received 4 weeks PEG/R lead-in, 32 weeks BOC/PEG/R followed by 12 weeks PEG/R. Treatment Experienced Patients Late Responders The EPAR states that there is no evidence for added benefit of boceprevir beyond 32 weeks, therefore the same treatment regimen used in RESPOND-2 is recommended in the SPC for late responders who have received prior therapy (4 weeks PEG/R, 32 weeks BOC/PEG/R, 12 weeks PEG/R). Cirrhotic (Metavir F4) Patients. The CHMP highlighted that patients with cirrhosis represent a special case, being the group in whom achieving SVR may be expected to result in the most immediate clinical benefit, and that few cirrhotic patients were included in the boceprevir phase III trials. The CHMP recommend a conservative, 48 week treatment duration (4 weeks PEG/R, 44 weeks BOC/PEG/R) for these patients to optimise the likelihood of response.

19 In making this decision to recommend a different treatment duration for cirrhotic patients than for patients with all other degrees of liver fibrosis the CHMP created the subgroups F0-3 and F4, which were used in our submission. Null Responders Null responders (defined as patients with <2 log10 decline in HCV-RNA at TW 12 of their previous treatment regimen) were not included in the boceprevir phase III trials in treatment experienced patients. The efficacy of boceprevir triple therapy in this group is estimated based on the efficacy in patients who had a <1 log10 decrease in HCV-RNA at TW 4 of the trial treatment, and supported by data for those who had a <0.5 log10 decrease at TW4. The CHMP concluded that "there is hardly any doubt that boceprevir increases SVR rates in null responders, though an exact estimate of the magnitude of this effect is not available". The medical need in this patient group is recognised and a conservative 48 week treatment duration is recommended (4 weeks PEG/R, 44 weeks BOC/PEG/R). Stopping Rules The CHMP noted that the stopping rules were different in the phase III clinical trials and disputed the non-conservative stopping rule at 24 weeks in the SPRINT-2 trial. The following futility rule was proposed: discontinue all three drugs if the patient has HCV- RNA 100 IU/ml at TW 12; discontinue all three drugs if the patient has confirmed, detectable HCV-RNA at TW 24. These stopping rules allow assessment at the earlier time point of 12 weeks, rather than 24 weeks, ensuring that only patients with very low (or undetectable) HCV-RNA remain on treatment beyond week 12 and, as stated in the EPAR, they "simplify the posology of Victrelis because the same futility rule is used for both treatment naïve and previous treatment failure patients, and because the TW 12 and 24 time points are already part of the standard of care for monitoring HCV RNA testing during therapy with peginterferon and ribavirin." "The futility rule is based on the observations in the Phase 3 program that patients with HCV RNA levels 100 IU/mL at TW 12 are unlikely to achieve SVR; and patients with low levels of detectable HCV RNA at TW12 still had a substantial possibility of achieving SVR." Differences in Subgroups Based on Fibrosis Level As mentioned previously, the subgroups F0-3 and F4 were created due to the decision made by the EMA to recommend different treatment strategies for patients with cirrhosis (Metavir F4). Therefore in our submission we used the groups F0-3 and F4 to reflect the patient populations specified by the licence.

20 The ERG specifically asked for clarification on how the data for the subgroup F0-3 was pooled. The clinical study reports provide data by individual Metavir score (only the publications of the clinical trials used the F0-2 and F3-4). In our submission, for the subgroup F0-3, we included data for all patients with Metavir F0, F1, F2 or F3 and used an average across all patients. No weighting was used. All of the decisions around changes recommended by EMA should be considered conservative in that they mandate additional treatment compared with that used for the particular sub-groups within the trial. As such the efficacy values derived from the trial data when applied to the 'label' populations should also be regarded as conservative in that the additional treatment will, if anything, increase efficacy. B4a. The probabilities of sustained virologic response (SVR) reported in Tables B33 and B35 (pages 162 and 177) do not seem to correspond to the values reported for the SOC arms by Mills et al nor the ones presented in Table B21 (page 95). The values reported in Table B21 are the correct values and should have been incorporated into the final version of the model, see below: Efficacy TN Non black TN Black TE SVR_F4_SOC SVR_NR_SOC SVR_NR_LB Variable Probability of SVR in the PEG/R 48 F4 Probability of SVR in the PEG/R 48 NR Probability of SVR in the BOC PEG/R 48 NR The correct values have been applied in the model and the updated base case results for the treatment experienced population are presented below: Technology* (and comparators) Total discounted Total discounted Incremental cost Incremental QALY ICERs versus baseline (A) Incremental analysis cost QALY F0-F3-SOC 30, F0-F3-Label* 36, ,327 NA F4-SOC 49, , F4-Label* 49, NR-SOC 30,351 32,221 NR-Label* 37,122 39,834 SOC (All Metavir score levels 32,861 excludes 'null 32,606 responders') , ,772 7, ,977 4, NA NA

21 Label (All Metavir score levels - excludes 'null responders') (Comparisons with SOC)* 38, ,478 5, QALY, quality-adjusted life year; ICERs, incremental cost-effectiveness ratios 2,744 2,909 NA The ICER vs. the original baseline increases by 780 for 'F4-Label' patients, 830 for 'NR- Label' patients and 165 for all responder patients, F0-4.

22 B4b. Also, it is not clear how the SVR estimates for the Label arms were derived as they do not seem to have been reported in the clinical section of the manufacturer s submission (section 5.5, pages 73-93). Please clarify. The difference between the SVR estimates reported in the clinical section of the submission and the clinical inputs applied in the model, derive from the following points: Regarding the treatment naïve cohorts. In the model the SVR estimates are reported separately for Blacks and Non-Blacks while in the tables B18, B19 and B20 (pages 92-93) the estimates are reported for the total cohort. Additionally, in the same tables of the clinical section, the results are reported per the patients' response at week 8 (detectable and undetectable HCV-RNA) while in the model no distinction was made on the basis of the week 8 response. The label introduced an additional stopping rule to that of the trial setting one, in the treatment regimes of both treatment naïve and treatment experienced patients. i.e. for the treatment naïve treatment schedule an additional rule at week 12 was applied and for the treatment experienced treatment regimes an additional rule at week 24 was applied. Because of these differences in the stopping rules, an adjustment was necessary in order to reflect the impact of their application on the number of patients that achieved SVR. Consequently, in order to reflect the label, patients who do not meet the requirements to be considered as successful according to the additional stopping rule introduced by the label, were considered to have failed treatment. These patients were therefore subtracted from the final number of patients having achieved SVR (even where they did achieve SVR within the clinical trial). These patients have therefore only been included in the numerator and not the denominator of the calculation for the estimation of the SVR probabilities. The source of each of the efficacy inputs applied in the model is as follows: Efficacy TN TN Black TE Source Non black SVR_SOC Mills et al SVR_LB SPRINT-2 (199/292) 1 (20/45) 1 (85/132) 2 2 RESPOND-2 SVR_F4_SOC * SPRINT-2 (5/13) 1 (5/13) 1 2 Table B21 SVR_F4_LB * SPRINT-2 (10/24) 1 (10/24) 1 2 Table B21 SVR_NR_SOC Table B21 SVR_NR_LB Table B21 SVR_SOC Mills et al *Assumed to be the same as Non-Black due to small number of subjects MSD. Registered Office Hertford Road, Hoddesdon, Hertfordshire EN!! 9BU Registered in England No

23 The derivation of the efficacy inputs above is provided below: Treatment Naïve patients, F0-F3 Label arm For this analysis, the patient group of the Response Guided Therapy arm of SPRINT- 2 were utilised. This was done in order to reflect the treatment effect of the 24 week treatment duration with boceprevir. In particular, among the 292 Non-black patients, 202 achieved SVR. However, 3 were excluded for failing the week 24 futility rule, resulting in a rate of 199/292. Among the 45 Black patients, 20 achieved SVR (none were excluded from the SVR count due to futility at week 12 or 24). Control arm The SVR reported by Mills et al was applied in the TN F0-F3 control arm. Treatment Naïve patients, F4 For this analysis, the Fixed Duration arm of SPRINT-2 patients were utilized in order to reflect the treatment effect of 44 week treatment with boceprevir. Label arm A pooled analysis was used for Blacks (n=2), Non-Black (n=24). Among the 24 patients, 10 achieved SVR (non were excluded from the SVR count due to futility at week 12 or 24) Control arm Among the 13 patients, 6 patients achieved SVR. However, 1 SVR patient failed the week 12 futility rule and was excluded, resulting in an SVR rate of 5/13. There were no Black individuals in this subgroup, so the same rate was applied for both groups. Treatment Experienced patients, F0-F3 For this analysis, the patient group of the Response Guided Therapy arm of RESPOND-2 were utilized. This was done in order to reflect the treatment effect of the 36 week treatment duration with boceprevir. Control arm The SVR reported by Mills et al was applied in the TE F0-F3 control arm Label arm Among 132 patients, 85 achieved SVR (none were excluded from the SVR count due to futility at week 12 or 24). MSD. Registered Office Hertford Road, Hoddesdon, Hertfordshire EN!! 9BU Registered in England No

24 Treatment Experienced patients, F4 Label arm In the label arm of the TE, F4, the SVR estimated by the Meta-analysis of the Proportion Treatment Experienced Patients with Cirrhosis and Prior Null Responders Achieving SVR (reported in table B21 of the submission) was applied. Control arm In the label arm of the TE, F4, the SVR estimated by the Meta-analysis of the Proportion Treatment Experienced Patients with Cirrhosis and Prior Null Responders Achieving SVR (reported in table B21 of the submission) was applied. Treatment Experienced patients, Null responders In the label arm of the TE, F4, the SVR estimated by the Meta-analysis of the Proportion Treatment Experienced Patients with Cirrhosis and Prior Null Responders Achieving SVR (reported in table B21 of the submission) was applied. B5. In table B35 (pages ), reference is made to a meta-analysis in Appendix 15. Please provide this meta-analysis as it is not included (the supplied Appendix 15 contains markov traces). The Meta-analysis has been provided as a reference: Mills et al, 2011 (reference 38). Therefore, "Mills et al, 2011" should be substituted for "Appendix 15" in Table 33. B6. Please provide the derivation of the estimates presented on Table B35 (page182) for the cost of monitoring people while they are on different treatment regimens cm_pr, cm_rgt, cm_prb as this is not clear in section of the manufacturer s submission. These are average overall monitoring cost for the different treatment regimens which were initially included in the model. They were inadvertently not removed when the decision was later made that the use of weekly costs was the more appropriate approach (reported in section of the submission). These redundant parameters are not active in the final version of the model used for the development of this economic analysis, therefore, the variables cm_pr, cm_rgt, cm_prb should be ignored. B7. Table B33 (page 161) reports different values for the proportions of anaemic patients who used erythropoietin (EPO) from the ones presented in Table B35 (page 176). Please justify these differences, presenting the sources and the rationale for the use of the different sets of values in the analysis. The proportions described in table B33 are those used in the modelling. The rationale for the selection of these figures is provided on p.114 of the original submission as below: MSD. Registered Office Hertford Road, Hoddesdon, Hertfordshire EN!! 9BU Registered in England No

25 "Management options for treatment-related anaemia in England and Wales include dose reduction of ribavirin and/or administration of erythropoietin. Erythropoietin is not currently licensed in the UK for use in patients with anaemia related to HCV therapy, but despite this some erythropoietin is used for this indication in the UK. In the boceprevir clinical trials SPRINT-2, RESPOND-2 and P05685 anaemia management was the same as for clinical practice in England and Wales. For patients whose haemoglobin fell to less than 10g/dL ribavirin dose reduction and/or use of erythropoietin was allowed. The exception being that as erythropoietin was a) provided at no cost and b) written into the trial protocol, significantly more use was seen than would be expected in routine clinical use in England and Wales. Feedback from an advisory board of 7 senior healthcare professionals who have experience in managing anaemia among patient receiving treatment for chronic hepatitis C in England and Wales, suggested that the extent of erythropoietin use varies from no use to use in almost half of cases, with the highest use of erythropoietin reported in more specialised HCV centres. It is anticipated that the main management strategy for anaemia in patients receiving boceprevir will involve dose reduction of ribavirin, depending on the severity of anaemia. In the pooled safety analysis of three of the boceprevir clinical trials, ribavirin dose reduction was used to manage anaemia in 42% of cases in PEG2b/R treated patients and 53% of cases in BOC/PEG2b/R treated patients, with or without concomitant erythropoietin. There is evidence to support that temporary modification in ribavirin dose does not have an impact on overall success of PEG/R regimens. In the boceprevir clinical trials, the percentage of patients in the combined BOC/PEG/R arms with anaemia who were managed by ribavirin dose reduction, erythropoietin, both or neither achieving SVR was in the range of 71% to 81% (SPRINT-2) and 72% to 83% (RESPOND-2), which supports the concept that treatment associated anaemia does not negatively influence SVR, and in fact may indicate eventual treatment success 16. From this post-hoc analysis it appears that ribavirin dose reduction and/or erythropoietin use do not impact on final results, in terms of SVR rates. This is in line with findings from a post-hoc analysis of the IDEAL study, which found that in patients experiencing anaemia on PEG/R combination therapy, ribavirin dose reduction was not associated with lower SVR rates 16." However the submission omits the rationale for selection of 0.25 as the consistent value across all patient groups. This value was selected given that feedback from the MSD. Registered Office Hertford Road, Hoddesdon, Hertfordshire EN!! 9BU Registered in England No

26 clinicians cited above that suggested 0.2 would be an overall absolute maximum value. This was rounded up to 0.25 to provide a conservative estimate given the uncertainty around the actual usage that will occur in clinical practice. The values listed in table B35 are the 'within trial' actual proportions, were not used in the analyses and this should have been noted as such in the submission. MSD. Registered Office Hertford Road, Hoddesdon, Hertfordshire EN!! 9BU Registered in England No