Addendum to Commission A14-12 (canagliflozin) 1

Transcription

1 IQWiG Reports Commission No. A14-24 Addendum to Commission A14-12 (canagliflozin) 1 Addendum Commission:A14-24 Version: 1.0 Status: 4 August Translation of addendum A14-24 Addendum zum Auftrag A14-12 (Canagliflozin) (Version 1.0; Status: 4 August 2014). Please note: This translation is provided as a service by IQWiG to English-language readers. However, solely the German original text is absolutely authoritative and legally binding.

2 Publishing details Publisher: Institute for Quality and Efficiency in Health Care Topic: Addendum to Commission A14-12 (canagliflozin) Commissioning agency: Federal Joint Committee Commission awarded on: 22 July 2014 Internal Commission No.: A14-24 Address of publisher: Institute for Quality and Efficiency in Health Care Im Mediapark 8 (KölnTurm) Cologne Germany Tel.: +49 (0) Fax: +49 (0) berichte@iqwig.de Internet: Institute for Quality and Efficiency in Health Care (IQWiG) - i -

3 IQWiG employees involved in the addendum 2 : Beate Wieseler Lars Beckmann Regine Potthast Keywords: canagliflozin, diabetes mellitus type 2, benefit assessment 2 Due to legal data protection regulations, employees have the right not to be named. Institute for Quality and Efficiency in Health Care (IQWiG) - ii -

4 Table of contents Page List of tables... iv List of figures... v List of abbreviations... vi 1 Background Assessment of the DIA3009 study: canagliflozin versus glimepiride Study characteristics Results References Appendix A Supplementary analyses Institute for Quality and Efficiency in Health Care (IQWiG) - iii -

5 List of tables Page Table 1: Characteristics of the study population RCT, direct comparison treatment regimen canagliflozin vs. treatment regimen glimepiride (DIA3009 study, target population a, without countries with approval of 8 mg glimepiride b )... 3 Table 2: Characteristics of the study population RCT, direct comparison treatment regimen canagliflozin vs. treatment regimen glimepiride (DIA3009 study, total population)... 4 Table 3: Results RCT, direct comparison treatment regimen canagliflozin + metformin vs. treatment regimen glimepiride (DIA3009 study, target population, without countries with approval of 8 mg glimepiride, if no analysis was available for this population, an analysis of the target population is shown in which patients were censored when reaching 8 mg glimepiride; data at week 104) Table 4: Results RCT, direct comparison treatment regimen canagliflozin + metformin vs. treatment regimen glimepiride (DIA3009 study, target population, without countries with approval of 8 mg glimepiride): genital mycosis, subgroups by sex Institute for Quality and Efficiency in Health Care (IQWiG) - iv -

6 List of figures Page Figure 1: Change in HbA1c in the course of the DIA3009 study (target population, without countries with approval of 8 mg glimepiride, mixed-effects model repeated measures [MMRM] [no information on the proportion of missing values available])... 6 Figure 2: Proportion of patients with HbA1c < 6.5% in the course of the first year of the DIA3009 study (total population, last observation carried forward [LOCF] [3])... 7 Figure 3: Time to first hypoglycaemia (symptomatic and documented [< 70 mg/dl]) in the DIA3009 study (target population, without countries with approval of 8 mg glimepiride)... 8 Figure 4: Mean number of symptomatic and documented hypoglycaemias in the 3 treatment groups in the course of the DIA3009 study (target population, without countries with approval of 8 mg glimepiride)... 9 Figure 5: Mean number of symptomatic and documented hypoglycaemias under different dosages of glimepiride in the course of the DIA3009 study (target population, the data of patients with a glimepiride dose of > 6 mg [dose level 5] were censored); level 1: 1 mg, level 2: 2 mg, level 3: 4 mg, level 4: 6 mg glimepiride Figure 6: Interaction test for sex for the outcome "genital mycosis", canagliflozin 100 mg/day vs. glimepiride in the DIA3009 study (target population, without countries with approval of 8 mg glimepiride). RR with correction of 0.5 in each cell Figure 7: Interaction test for sex for the outcome genital mycosis, canagliflozin 300 mg/day vs. glimepiride in the DIA3009 study (target population, without countries with approval of 8 mg glimepiride). RR with correction of 0.5 in each cell Institute for Quality and Efficiency in Health Care (IQWiG) - v -

7 List of abbreviations Abbreviation CrCl egfr EQ-5D G-BA HbA1c IQWiG IWQoL MMRM SF-36 Meaning creatinine clearance estimated glomerular filtration rate European Quality of Life-5 Dimensions Gemeinsamer Bundesausschuss (Federal Joint Committee) haemoglobin A1c Institut für Qualität und Wirtschaftlichkeit im Gesundheitswesen (Institute for Quality and Efficiency in Health Care) Impact of Weight on Quality of Life-Lite mixed-effects model repeated measures Short Form (36) Health Survey Institute for Quality and Efficiency in Health Care (IQWiG) - vi -

8 1 Background On 22 July 2014 the Federal Joint Committee (G-BA) commissioned the Institute for Quality and Efficiency in Health Care (IQWiG) to conduct a supplementary assessment for Commission A14-12 (benefit assessment of canagliflozin [1]). In the commenting procedure on the assessment of canagliflozin, in its comment from 7 July 2014 and in the oral hearing conducted on 22 July 2014, the pharmaceutical company (hereinafter abbreviated to the company ) submitted supplementary information to the G-BA that went beyond the information in the dossier [2]. These refer to data on the DIA3009 study (comparison of canagliflozin [100 mg and 300 mg] versus glimepiride, each in combination with metformin). The G-BA commissioned IQWiG with a comparison of the results of the patient-relevant outcomes of canagliflozin with metformin versus glimepiride with metformin from the DIA3009 study (therapeutic indication B, in combination with metformin) under consideration of the documents subsequently submitted. The responsibility for the present assessment and the results of the assessment lies exclusively with IQWiG. The assessment is forwarded to the G-BA. The G-BA decides on the added benefit. Institute for Quality and Efficiency in Health Care (IQWiG) - 1 -

9 2 Assessment of the DIA3009 study: canagliflozin versus glimepiride + metformin The present addendum supplements the dossier assessment of canagliflozin with additional information on blood-glucose lowering and on the course of hypoglycaemias in the DIA3009 study, which is mainly based on the company s comment. Moreover, results of the treatment groups on patient-relevant outcomes are compared. Where possible, an analysis submitted with the comment on the dossier assessment is used for this comparison [2]. This analysis comprises the target population defined by the company according to the German approval status under exclusion of patients from countries in which glimepiride is approved at a dosage of 8 mg daily. Randomization is maintained after this exclusion because randomization of the patients in the DIA3009 study was stratified by countries. Hence this analysis has a lower risk of bias with regard to the exclusion of patients with 8 mg glimepiride than the analysis presented in the dossier, in which patients were censored on reaching the 8 mg dosage, which is not approved in Germany. 2.1 Study characteristics Study design A comprehensive description of the design of the DIA3009 study including the interventions used (canagliflozin 100 and 300 mg at a fixed dosage and glimepiride at variable dosage with titration to a target value, each in combination with metformin) can be found in Section of the dossier assessment on canagliflozin [1]. Study population Table 1 shows the characteristics of the target population of the DIA3009 study, i.e. the population which fulfilled important criteria of the German approval status. The company defined the target population as those patients who received no loop diuretics, had an estimated glomerular filtration rate (egfr) 60 ml/min/1.73 m 2 or a creatinine clearance (CrCl) 60 ml/min on the day of randomization, received a maximum metformin dose of 3000 mg and no retard formulation of metformin (not approved in Germany) until the study visit on the day of randomization. In addition to this limitation, in the commenting procedure the company presented an analysis that did not consider patients from countries in which 8 mg glimepiride are approved. As far as possible, this analysis is used for the presentation of the results of the outcomes. Information not presented by the company in this analysis was taken from other analysis populations: on the one hand from the total population of the study, and on the other from an analysis of the target population, in which patients were censored at a dosage of > 6 mg glimepiride (respective information is provided). Table 2 shows the characteristics of the total population of the DIA3009 study. Institute for Quality and Efficiency in Health Care (IQWiG) - 2 -

10 Table 1: Characteristics of the study population RCT, direct comparison treatment regimen canagliflozin vs. treatment regimen glimepiride (DIA3009 study, target population a, without countries with approval of 8 mg glimepiride b ) Group canagliflozin 100 mg canagliflozin 300 mg glimepiride N a Age [years]: mean (SD) 56.5 (8.8) 56.3 (8.9) 56.1 (8.6) Sex F/M [%] 47/53 56/44 41/59 Disease duration [years]: mean (SD) 4.4 (3.9) 5.8 (5.7) 5.6 (4.1) HbA1c at start of study [%]: mean (SD) 7.6 (0.8) 7.7 (0.7) 7.7 (0.8) HbA1c at start of study [%]: categories [n (%)] < 7.0% ND ND ND 7.0% to < 8.0% ND ND ND 8.0% to < 9.0% ND ND ND 9.0% to 10% ND ND ND > 10% ND ND ND BMI [kg/m 2 ]: mean (SD) 31.3 (4.6) 31.1 (4.8) 31.4 (5.4) Body weight [kg]: mean (SD) 87.4 (17.1) 86.4 (17.7) 88.3 (17.3) a: The company defined the target population as those patients who received no loop diuretics, had an estimated glomerular filtration rate (egfr) 60 ml/min/1.73 m 2 or a creatinine clearance (CrCl) 60 ml/min on the day of randomization, received a maximum metformin dose of 3000 mg and no retard formulation of metformin (not approved in Germany) until the study visit on the day of randomization; moreover, patients from countries in which 8 mg glimepiride is approved were excluded from this population. b: The population comprised patients from the following countries; Bulgaria, Denmark, Finland, Germany, India, Norway, Poland, Romania, Russia, Slovakia, Ukraine. BMI: body mass index; F: female; HbA1c: haemoglobin A1c; M: male; N: number of patients in the subpopulation; ND: no data; RCT: randomized controlled trial; SD: standard deviation; vs.: versus Institute for Quality and Efficiency in Health Care (IQWiG) - 3 -

11 Table 2: Characteristics of the study population RCT, direct comparison treatment regimen canagliflozin vs. treatment regimen glimepiride (DIA3009 study, total population) Group canagliflozin 100 mg canagliflozin 300 mg glimepiride N a Age [years]: mean (SD) 56.2 (9.4) 55.6 (9.2) 56.1 (9.0) Sex F/M [%] 48/52 50/50 45/55 Disease duration [years]: mean (SD) 6.5 (5.5) 6.8 (5.5) 6.5 (5.0) HbA1c at start of study [%]: mean (SD) 7.8 (0.8) 7.8 (0.8) 7.8 (0.8) HbA1c at start of study [%]: categories [n (%)] b < 7.0% 59 (12.2) 71 (14.6) 65 (13.5) 7.0% to < 8.0% 235 (48.7) 223 (46.0) 225 (46.7) 8.0% to < 9.0% 147 (30.4) 147 (30.3) 147 (30.5) 9.0% to 10% 40 (8.3) 43 (8.9) 42 (8.7) > 10% 2 (0.4) 1 (0.2) 3 (0.6) BMI [kg/m 2 ]: mean (SD) 31.0 (5.3) 31.2 (5.4) 31.0 (5.6) Body weight [kg]: mean (SD) 86.9 (20.0) 86.7 (19.4) 86.7 (19.9) a: Patients who took at least one dose of medication and for whom 1 recording after the start of the study was available for HbA1c. b: Information on all randomized patients who took at least one dose of medication (mitt): canagliflozin 100 mg : 483, canagliflozin 300 mg : 485, glimepiride : 482 [3]. BMI: body mass index; F: female; HbA1c: haemoglobin A1c; M: male; mitt: modified intention to treat; N: number patients; ND: no data; RCT: randomized controlled trial; SD: standard deviation; vs.: versus No relevant differences between treatment groups regarding age, disease duration or HbA1c value at the start of the study were evident for the target population or for the total population. The patients in the target population had a mean age of about 56 years. The proportion of women was about 50% in the canagliflozin groups and about 40% in the glimepiride group. The disease duration was 4 to 6 years. HbA1c (long-term marker for the average blood glucose level) in the target population had a mean value of about 7.7% at the start of the study. In the total population (no corresponding data were available for the target population), the HbA1c value was below < 7.0% in 12% to 15% of the patients. According to current knowledge, for these patients one cannot assume inadequate glycaemic control that would have required intensified therapy. Particularly in these patients, intensified blood-glucose lowering therapy was associated with an increased risk of hypoglycaemia. Institute for Quality and Efficiency in Health Care (IQWiG) - 4 -

12 Summary In the DIA3009 study, there were relevant differences between the treatment arms with regard to the specified target blood glucose levels and the therapeutic strategies determined by them. In the canagliflozin arms of the study, target blood glucose levels were not aimed at ( titration to target levels was performed without dose changes and merely to maintain blinding) and fixed dosage was used. In the glimepiride arm, in contrast, titration was specified by an algorithm and orientated towards near-normal target levels. The additional analysis for the target population, but without patients from countries in which a daily dosage of 8 mg glimepiride was approved, which was presented in the comment on the dossier assessment, did not solve the key problem of the study: It still remained unclear from the consideration of the results on patient-relevant outcomes whether the observed effects were due to the drugs used or to the therapeutic strategy. The results of the DIA3009 study on HbA1c and hypoglycaemias illustrate the problem described above. This is explained in the following section. 2.2 Results Blood-glucose lowering: HbA1c The following Figure 1 shows the mean change in HbA1c compared with the start of the study during the 104-week treatment phase of the DIA3009 study in the target population. Institute for Quality and Efficiency in Health Care (IQWiG) - 5 -

13 Figure 1: Change in HbA1c in the course of the DIA3009 study (target population, without countries with approval of 8 mg glimepiride, mixed-effects model repeated measures [MMRM] [no information on the proportion of missing values available]) The time course of the change in HbA1c showed a rapid decrease in HbA1c under the targetlevel directed treatment with glimepiride during the first weeks of the study; the maximum change in HbA1c was reached after 18 weeks. A decrease in HbA1c was also observed in the canagliflozin groups. However, in the first phase of the study this was less pronounced than in the glimepiride group. In the first 3 to 6 weeks of the study the difference in the decrease in HbA1c was not yet markedly pronounced. This can be explained by the fact that HbA1c is a long-term marker that shows an average blood-glucose level during a period of 6 to 12 weeks. It can therefore not be expected that the effectiveness of intensified treatment can be evaluated in the first weeks by means of HbA1c. The difference between treatment groups was greatest after 18 weeks. At first the changes in HbA1c value approximated each other in the course of the study in all treatment groups; at the Institute for Quality and Efficiency in Health Care (IQWiG) - 6 -

14 end of the study the decrease in HbA1c value was more pronounced under canagliflozin + canagliflozin than under glimepiride. Figure 2 shows the proportion of patients in the total population of the study who achieved an HbA1c < 6.5% in the course of the first year of the study. % patients Week 8 Week 12 Week 18 Week 26 Week 36 Week 44 Week 52 Canagliflozin 100 mg Canagliflozin 300 mg Glimepiride Figure 2: Proportion of patients with HbA1c < 6.5% in the course of the first year of the DIA3009 study (total population, last observation carried forward [LOCF] [3]) The proportion of patients with HbA1c below 6.5% in the total population of the study increased more steeply in the glimepiride group and reached a higher value than in the canagliflozin groups. At week 18, the difference was 19 percentage points (38% versus 19%) between glimepiride and canagliflozin 100 mg, and 13 percentage points (38% versus 25%) between glimepiride and canagliflozin 300 mg. Hence there was a higher risk of hypoglycaemia in the glimepiride group than in the canagliflozin groups. There were no data on the proportion of patients with an HbA1c value below 6.5% for the target population. It can therefore not be assessed whether the differences are smaller without the patients who received 8 mg glimepiride in the course of the study. Hypoglycaemias The time course of the occurrence of hypoglycaemia corresponds to the described course of blood-glucose lowering. Figure 3 shows the time to occurrence of a first hypoglycaemic event (Kaplan-Meier curve) in the target population of the DIA3009 study (without patients from countries in which a daily dosage of 8 mg glimepiride is approved). The curves take into account the hypoglycaemias that were symptomatic and documented (blood glucose level Institute for Quality and Efficiency in Health Care (IQWiG) - 7 -

15 < 70 mg/dl). It is pointed out that the value of < 70 mg/dl is a low threshold for documented hypoglycaemia. The mean number of symptomatic and documented hypoglycaemias in the 3 treatment groups in the course of the study is presented in Figure 4. Figure 5 shows the course of the mean number of symptomatic and documented hypoglycaemias in the different dosing steps of glimepiride (target population, censored at > 6 mg). Figure 3: Time to first hypoglycaemia (symptomatic and documented [< 70 mg/dl]) in the DIA3009 study (target population, without countries with approval of 8 mg glimepiride) Institute for Quality and Efficiency in Health Care (IQWiG) - 8 -

16 Figure 4: Mean number of symptomatic and documented hypoglycaemias in the 3 treatment groups in the course of the DIA3009 study (target population, without countries with approval of 8 mg glimepiride) Institute for Quality and Efficiency in Health Care (IQWiG) - 9 -

17 Figure 5: Mean number of symptomatic and documented hypoglycaemias under different dosages of glimepiride in the course of the DIA3009 study (target population, the data of patients with a glimepiride dose of > 6 mg [dose level 5] were censored); level 1: 1 mg, level 2: 2 mg, level 3: 4 mg, level 4: 6 mg glimepiride For the majority of patients with hypoglycaemia in the glimepiride group, the first hypoglycaemia was documented in the first study phase, in which titration of glimepiride took place (Figure 3). The mean number of hypoglycaemias over the total course of the study was higher in the glimepiride group than in the canagliflozin groups (Figure 4). The hypoglycaemias were mainly documented in patients with a daily dosage of 6 mg glimepiride (Figure 5). It therefore remains unclear whether and to which extent this increased number of hypoglycaemias was caused by the drug glimepiride or by the therapeutic strategy of titration to near-normal target levels, which was only applied in the glimepiride group. Figure 5 also shows that the hypoglycaemias primarily occurred in the titration phase and under a dosage of 6 mg glimepiride, which underlines the influence of the therapeutic strategy on the results on hypoglycaemias. The proportion of patients in the glimepiride group for which repeated hypoglycaemias were documented also indicates intensive blood glucose lowering (2 symptomatic and documented hypoglycaemias in 5 of 43 patients with hypoglycaemias, 3 hypoglycaemias in 27 of 43 patients with hypoglycaemias). Under canagliflozin, there was only one patient with more than one symptomatic and documented hypoglycaemia (canagliflozin 100 mg group: 3 hypoglycaemias in 1 of 4 patients with hypoglycaemias). It remains unclear whether the repeated occurrence of hypoglycaemias is associated with the dosage of glimepiride because there were no corresponding data. Institute for Quality and Efficiency in Health Care (IQWiG)

18 Finally, it remains unclear from the results of the DIA3009 study whether and to which extent the observed differences in the results between canagliflozin and glimepiride + metformin were caused by the drugs or by the different therapeutic strategies in the treatment groups. Further outcomes The DIA3009 study was not designed to infer an advantage or non-inferiority of canagliflozin versus glimepiride for the outcomes mortality, cardiovascular events or cerebral events, which are particularly relevant to the area of treatment. Due to the above-described deficiencies in study design, such data would however not be interpretable in the sense of an advantage specific to one of the drugs. Table 3 compares the results in the treatment groups of the DIA3009 study. It is pointed out that the use of canagliflozin with an initial administration of 300 mg/day and the administration of 100 mg/day without the possibility of dose increase do not comply with the specifications of the Summary of Product Characteristics [1,4]. This can lead to misjudgements (both overestimation and underestimation) of the benefit and harm of canagliflozin. Mortality There were no data on mortality for the target population without patients from countries in which the daily dosage of 8 mg glimepiride is approved. Hence the results of an analysis of the target population were used in which patients who reached the glimepiride dosage of 8 mg were censored. There were no statistically significant differences between the treatment groups. Morbidity There were no evaluable data on morbidity. There was no possibility to use information from the recording of adverse events for the assessment of cardiovascular or cerebral events because the company presented no analysis for these outcomes that excluded patients with a glimepiride dosage of 8 mg (or in which these were censored when reaching this dosage). The data of the total population of the study were unsuitable because of the high proportion of patients who received 8 mg glimepiride because the bias caused by the unapproved high dosage cannot be estimated. Health-related quality of life The company used 3 instruments to measure health-related quality of life in the study (Short Form (36) Health Survey [SF-36], Impact of Weight on Quality of Life-Lite [IWQoL-Lite], European Quality of Life-5 Dimensions [EQ-5D]). In the comment, it presented no analyses for these questionnaires for the target population without countries with approval of 8 mg glimepiride. However, in the dossier it presented an analysis in which patients were censored when reaching the glimepiride dosage of 8 mg (primary analysis). Data for the total study Institute for Quality and Efficiency in Health Care (IQWiG)

19 duration were only available for the SF-36. However, too few patients were considered in the analysis of the SF-36 at week 104 (canagliflozin 100 mg : 58%; canagliflozin 300 mg : 54%; glimepiride : 35%). Hence no evaluable data were available for the time point of 104 weeks. In general, no statistically significant differences between the treatment groups were observed in the primary analysis of the data on all 3 questionnaires (at the time points of 52 and/or 104 weeks). Adverse events The data on hypoglycaemias were not interpretable for the reasons described above. The results on adverse events, serious adverse events and treatment discontinuations due to adverse events were also not interpretable, especially as hypoglycaemias were also recorded under these outcomes. The statistically significant effect to the disadvantage of canagliflozin 300 mg for treatment discontinuations due to adverse events cannot be interpreted even under consideration of the direction of the bias with regard to hypoglycaemias (in favour of canagliflozin) because the initial dosage of 300 mg does not concur with use according to the Summary of Product Characteristics. For genital mycosis, there were large effects to the disadvantage of both canagliflozin dosages, i.e. also under canagliflozin 100 mg, although this fixed dosage presumably underestimates the risk. The company presented genital mycosis separately for men and for women in its comment. This presentation was not followed because the data showed no effect modification by the factor sex (see Appendix A). Summary No added benefit of canagliflozin in comparison with glimepiride can be derived from the comparison of the results of the DIA3009 study. Institute for Quality and Efficiency in Health Care (IQWiG)

20 Table 3: Results RCT, direct comparison treatment regimen canagliflozin vs. treatment regimen glimepiride (DIA3009 study, target population, without countries with approval of 8 mg glimepiride, if no analysis was available for this population, an analysis of the target population is shown in which patients were censored when reaching 8 mg glimepiride; data at week 104) Outcome category outcome Mortality all-cause mortality a (104 weeks) Morbidity N Health-related quality of life Adverse events (104 weeks) Hypoglycaemia canagliflozin 100 mg Patients with events n (%) N canagliflozin 300 mg Patients with events n (%) N glimepiride Patients with events n (%) Canagliflozin 100 mg vs. glimepiride Effect estimate RR [95% CI] p-value (1.1) (0.7) (0.3) 3.02 [0.32; 28.87] No evaluable data available No evaluable data available Severe hypoglycaemia (0.7) (0.8) (2.0) 0.35 [0.04; 3.35] Symptomatic and documented hypoglycaemia (blood glucose 70 mg/dl) (2.9) (1.6) (29.1) 0.10 [0.04; 0.27] < Canagliflozin 300 mg vs. glimepiride Effect estimate RR [95% CI] p-value 1.97 [0.20; 18.98] b c 0.42 [0.06; 3.00] b c 0.05 [0.01; 0.22] < Acute renal failure (0.7) (0.8) [0.13; 77.18] d 3.44 [0.14; 83.68] d c c Genital mycosis e (7.9) (11.6) [1.45; ] d < c [2.15; ] d < c (continued) Institute for Quality and Efficiency in Health Care (IQWiG)

21 Table 3: Results RCT, direct comparison treatment regimen canagliflozin vs. treatment regimen glimepiride (DIA3009 study, target population, without countries with approval of 8 mg glimepiride, if no analysis was available for this population, an analysis of the target population is shown in which patients were censored when reaching 8 mg glimepiride; data at week 104) (continued) Outcome category outcome N canagliflozin 100 mg Patients with events n (%) N canagliflozin 300 mg Patients with events n (%) N glimepiride Patients with events n (%) Canagliflozin 100 mg vs. glimepiride Effect estimate RR [95% CI] p-value Canagliflozin 300 mg vs. glimepiride Effect estimate RR [95% CI] p-value 1.48 [0.57; 3.85] f 0.71 [0.30; 1.65] f Adverse events (104 weeks) Urinary tract infection (5.0) (7.0) (4.7) 1.06 [0.38; 2.94] f c c Neoplasia No evaluable data available g Overall rate AEs (53.6) (62) (56.8) Overall rate SAEs (6.4) (6.2) (8.8) 0.73 [0.32; 1.66] f c c Treatment discontinuations due to AEs (4.3) (10.9) (3.4) 1.27 [0.40; 4.06] f 3.21 [1.19; 8.68] f c c a: Analysis of the target population shown in which patients were censored when reaching 8 mg glimepiride. b: Institute s calculation, Peto OR. c: Institute s calculation, unconditional exact test (CSZ method according to [5]). d: Institute s calculation, RR with correction of 0.5 in each cell. e: The company presented genital mycosis separated by sex. This presentation was not followed because there was no effect modification by sex (interaction p = for canagliflozin 100 mg vs. glimepiride, and for canagliflozin 300 mg vs. glimepiride). f: Institute s calculation, RR. g: No data available in the target population. AE: adverse event; CI: confidence interval; CSZ: convexity, symmetry, z score; N: number of patients; n: number of patients with events; OR: odds ratio; RCT: randomized controlled trial; RR: relative risk; SAE: serious adverse event; vs.: versus Institute for Quality and Efficiency in Health Care (IQWiG)

22 3 References 1. Institut für Qualität und Wirtschaftlichkeit im Gesundheitswesen. Canagliflozin: Nutzenbewertung gemäß 35a SGB V; Dossierbewertung; Auftrag A14-12 [online]. 12 June 2014 [accessed: 31 July 2014]. (IQWiG-Berichte; Volume 225). URL: 2. Janssen-Cilag GmbH. Stellungnahme zum IQWiG-Bericht Nr. 225: Canagliflozin; Nutzenbewertung gemäß 35a SGB V; Dossierbewertung; Auftrag A [Soon available under: in the document "Zusammenfassende Dokumentation"]. 3. Janssen Research & Development. A randomized, double-blind, 3-arm parallel-group, 2- year (104-week), multicenter study to evaluate the efficacy, safety, and tolerability of JNJ mg and JNJ mg compared with glimepiride in the treatment of subjects with type 2 diabetes mellitus not optimally controlled on metformin monotherapy: the CANTATA-SU trial (CANagliflozin Treatment And Trial Analysis Sulfonylurea); 104- week double-blind treatment phase; DIA3009; study DIA3009; clinical study report [unpublished] Janssen. Invokana 100 mg Filmtabletten; Invokana 300 mg Filmtabletten: Fachinformation [online]. November 2013 [accessed: 31 July 2014]. URL: 5. Martín Andrés A, Silva Mato A. Choosing the optimal unconditioned test for comparing two independent proportions. Computat Stat Data Anal 1994; 17(5): Institute for Quality and Efficiency in Health Care (IQWiG)

23 Appendix A Supplementary analyses Canagliflozin 100mg/day vs. glimepiride Genital mycosis Random effects model - DerSimonian and Laird Study pool Study cana 100/met n/n glime/met n/n RR (95% CI) weight RR 95% CI Women DIA3009 8/66 0/ [0.93, ] Men DIA3009 3/74 0/ [0.43, ] All Total 11/140 0/ [1.49, 88.72] Heterogeneity: Q=0.10, df=1, p=0.752, I²=0% Overall effect: Z Score=2.35, p=0.019, Tau= favours cana 100/met favours glime/met Heterogeneity among study pools: Q=0.10, df=1, p=0.755, I²=0% Figure 6: Interaction test for sex for the outcome "genital mycosis", canagliflozin 100 mg/day vs. glimepiride in the DIA3009 study (target population, without countries with approval of 8 mg glimepiride). RR with correction of 0.5 in each cell. Canagliflozin 300mg/day vs. glimepiride Genital mycosis Random effects model - DerSimonian and Laird Study pool Study cana 300/met n/n glime/met n/n RR (95% CI) weight RR 95% CI Women DIA /72 0/ [1.39, ] Men DIA3009 2/57 0/ [0.37, ] All Total 15/129 0/ [1.76, ] Heterogeneity: Q=0.31, df=1, p=0.580, I²=0% Overall effect: Z Score=2.50, p=0.012, Tau= favours cana 300/met favours glime/met Heterogeneity among study pools: Q=0.28, df=1, p=0.599, I²=0% Figure 7: Interaction test for sex for the outcome genital mycosis, canagliflozin 300 mg/day vs. glimepiride in the DIA3009 study (target population, without countries with approval of 8 mg glimepiride). RR with correction of 0.5 in each cell. Institute for Quality and Efficiency in Health Care (IQWiG)

24 Table 4: Results RCT, direct comparison treatment regimen canagliflozin vs. treatment regimen glimepiride (DIA3009 study, target population, without countries with approval of 8 mg glimepiride): genital mycosis, subgroups by sex Outcome category outcome N Adverse events (104 weeks) Genital mycosis canagliflozin 100 mg Patients with events n (%) N canagliflozin 300 mg Patients with events n (%) N glimepiride Patients with events n (%) Canagliflozin 100 mg vs. glimepiride Effect estimate RR [95% CI] p-value Total (7.9) (11.6) [1.45; ] a < b Interaction p = Women 66 8 (12.1) (18.1) [0.93; ] a b, c Men 74 3 (4.1) 57 2 (3.5) [0.43; ] a b Canagliflozin 300 mg vs. glimepiride Effect estimate RR [95% CI] p-value [2.15; ] a < b Interaction p = [1.39; ] a < b 7.59 [0.37; ] a b a: Institute s calculation, RR with correction of 0.5 in each cell. b: Institute s calculation, unconditional exact test (CSZ method according to [5]). c: Institute s calculation, asymptotic. Discrepancy between p-value (exact) and CI (asymptotic) due to different calculation methods. CI: confidence interval; CSZ: convexity, symmetry, z score; N: number of patients; n: number of patients with events; RCT: randomized controlled trial; RR: relative risk; vs.: versus Institute for Quality and Efficiency in Health Care (IQWiG)