_______________________________________________________________ Supporting Affidavit to the First to Ninth Respondents' Answering Affidavit _______________________________________________________________ I, the undersigned JONATHAN BERNHARD LEVIN do hereby make oath and say: 1. I am a Senior Statistician (at the rank of Senior Specialist Scientist) at the Medical Research Council of South Africa, and a member of the clinical trials committee of the Medicines Control Council (MCC). 2. In my capacity as a member of the MCC, I have the written authority of the Director-General of Health, Dr Ayanda Ntsaluba (Ntsaluba) to disclose such information as was required by me in the exercise of my powers and performance of my functions in terms of the Medicines and Related Substances Control Act No 101 of 1965. 3. The facts deposed to herein are within my personal knowledge and are true and correct. To the extent that I have relied on information and advice provided to me by other parties, their confirmatory affidavits are filed herewith. To the extent that I make submissions outside my area of speciality and in respect of law or procedure, I have been guided by advice of the respondents' attorney of record and counsel in this matter and make such submission in reliance of such advice. 4. 4.1 I have attached my curriculum vitae marked annexure "LM1". As is evident from my curriculum vitae, I am experienced in the design of clinical trials and analysis of data from clinical trials. 4.2 I have the necessary qualifications and experience to be able to comment on the matters referred to below. 5. The efficacy of intrapartum and neonatal single dose Nevirapine (NVP) in the prevention of mother-to-child transmission (PMTCT) of the HIV was established by the HIVNET 012 randomised trial in Uganda. This trial randomly assigned mothers to receive 200mg NVP orally at onset of labour and 2mg/kg to babies within 72 hours of birth, or AZT 600mg orally to the mother at onset of labour and 300mg every 3 hours until delivery, and 4mg/kg orally twice daily to babies for 7 days after birth. The primary efficacy endpoints were the rates of HIV infection and HIV- free survival at 6-8 weeks and 14-16 weeks. In the results presented in the Lancet paper by Guay et al, the rates of HIV infection or death were at week 6-8 NVP 12.8% versus AZT 23.1% and at week 14-16 NVP 14.4% versus AZT 27.6%. This represents a relative improvement of 47% for NVP, which was statistically significant (P=0.00021) i.e. this difference was real and could not have arisen by chance. It is important to note that the Lancet results were published in the interests of public health at a time that not all babies had reached the 14-16 week follow up. 6. Results are available for 20 months follow-up. The rates of HIV infection or death were roughly 20% for NVP versus 30% for AZT. This translates as a relative reduction of 33%, and an absolute risk difference of 10% i.e. NVP would save 10 out of every 100 babies born to HIV-positive mothers who agreed to participate in the programme as compared to the AZT regime. This AZT regime has never been shown to be effective, but it is certainly reasonable to assume that it could not be worse than placebo. Hence, the effect of NVP compared to placebo would be slightly higher. 7. Thus, HIVNET 012 provides conclusive evidence of the efficacy of NVP – it could be argued from a public health point of view that the month 20 result is more relevant than the week 14-16 result. Thus, it would reduce the MTCT (or death) rate by one-third rather than by one-half. It has been assumed that the difference between the week 14-16 results and the month 20 result is purely due to the effect of transmission through breast feeding, but another difference in the results is that week 14-16 diagnosis of HIV was by qualitative PCR while the month 20 diagnosis was by ELISA. This point will be considered again later in this affidavit. 8. The SAINT study was then carried out in South Africa. The full title of the study was "A prospective randomized open label clinical trial to determine the efficacy of Nevirapine compared with a combination of ZDV and 3TC in decreasing the peripartum mother to child transmission of HIV". The primary outcome measure was the incidence of HIV transmission which occurred in the peripartum period between the HIV positive mother and the exposed infant. Infants who had positive HIV DNA or RNA PCR within 48 hours of birth were defined as having been infected during the intrauterine period and were excluded from the primary analysis. A secondary outcome variable was the overall transmission rate up to 6 weeks. It has been claimed that the SAINT study confirmed the results of HIVNET 012. This claim can be challenged on the following grounds: 8.1 SAINT was designed as a superiority trial not a non-inferiority trial and failed to prove superiority of NVP. 8.2 The choice of the PETRA B regimen as the comparator arm when the efficacy of this regimen has not been conclusively proven. 8.3 The choice of outcome measure and the related problem of sensitivity of qualitative PCR for testing for HIV. 9. I will now consider each of these points in detail. 9.1 Superiority versus non-inferiority 9.1.1 The classical clinical trial is a superiority trial of an active drug versus a placebo; formerly if there is sufficient evidence the null hypothesis of no difference is rejected and we can conclude that the active drug is better ("Efficacy"). HIVNET 012 was in this mould with the very short AZT arm acting as a surrogate placebo. 9.1.2 An alternative in situations where a placebo is not ethical is to conduct a non-inferiority study – in which a test drug is compared to a reference drug of known efficacy. There can be a number of reasons for doing this e.g. for an improved safety profile (COX2 inhibitors versus more traditional NSAIDS) or for cost reasons (such as in MTCT). In a non-inferiority trial the role of the null and alternative hypotheses are reversed and in addition a non-inferiority margin must be specified. The test is carried out in practice by calculating a confidence interval for the treatment effect and checking that the relevant limit of the confidence interval lies within the non-inferiority margin e.g. if it is known that the Thai regime reduces MTCT by 12% in absolute terms, the non-inferiority limit could be set at 6%; to conclude non-inferiority the upper limit of the confidence interval for the difference between Thai and test should not be greater than 6%. 9.1.3 In the case of SAINT the study was designed as a superiority trial – the sample size calculation was based on the assumption that the peripartum rate for the NVP arm would be half that of the PETRA B arm. 9.1.4 The estimated rates for peripartum transmission were 5.6% for NVP as against 3.6% for ZDV/3TC; the overall rates of HIV infection or death were 14.3% for NVP versus 12.9% for ZDV/3TC. 9.1.5 Thus, it has to be concluded that under the formal hypothesis testing approach that governs the analysis of clinical trials, and given that the trial was designed to show the superiority of NVP, there is statistically no evidence at all that NVP is efficacious. 9.1.6 To explain this, in following HIVNET 012, SAINT aimed to show that NVP was better than AZT/3TC and thus must also be better than placebo. According to the way in which clinical trials are formally analysed, since SAINT failed to demonstrate superiority to AZT/3TC there is nothing that can be concluded. This does not mean that NVP does not work – it merely means that SAINT failed to prove that it works. 9.1.7 It is, therefore, incorrect to say that SAINT confirms the finding of HIVNET 012. 9.1.8 The trial that should have been carried out (or should still be carried out) should be a non-inferiority trial to compare the Thai regime (or a longer AZT regimen such as the ACTG076 regime) against NVP – and this would require that a predetermined non-inferiority margin be set up. 9.2 The choice of PETRA B as a comparator arm and the sensitivity of qualitative PCR testing 9.2.1 Given that a formal analysis of the SAINT study fails to prove efficacy, weaker evidence of efficacy can be provided by an epidemiological type analysis, comparing the results of SAINT to historical controls. This was the basis of the claims made at the Durban conference that NVP reduces MTCT by about 1/3 from the 22% transmission rate observed in the Vitamin A study conducted by some of the same investigators. 9.2.2 However when one is doing such an analysis, one should also look at the results of the PETRA study, given that this was the only study to test the comparator AZT/3TC regimen (the PETRA B treatment arm). 9.2.3 The results are summarized below for 6 weeks and 18 months, giving the rates of HIV transmission or death. Treatment Week 6 Month 18 Petra A 9.2% 20.7% Petra B 12.6% 24.4% Petra C 18.4% 25.7% Placebo 19.2% 26.6% 9.2.4 There was a significant reduction in MTCT for Petra A versus placebo at week 6 and also for Petra B versus placebo at week 6. Note that on the basis of this, since NVP in SAINT did worse than Petra B, although the difference was not statistically significant, the relative improvement at week 6 would be at best about 33% - less than the figure of 50% that is claimed by the Applicants. 9.2.5 More importantly at month 18 the differences between the four arms are no longer statistically significant; and the relative gain for Petra B over placebo is very minimal (less than 10% in relative terms and only 2.2% in absolute terms). 9.2.6 Since NVP did worse than Petra B in SAINT, on the basis of this historical comparison, in operational terms NVP would be of very little value. 9.2.7 The true picture probably lies somewhere between these two historical comparisons i.e. NVP does reduce MTCT, but probably by considerably less than the relative reduction of 50% as claimed by the Applicants. 9.2.8 It has been argued that the discrepancy between the week 6 and month 18 results can be attributed to the effect of breastfeeding. However the following results from the two South African Petra sites for mothers who did not breast feed reveal Treatment Week 6 Month 18 Petra A 9/89 (10.1%) 12/89 (13.5%) Petra B 8/91 (8.8%) 19/91 (20.9%) Petra C 15/87 (17.2%) 15/87 (17.2%) Placebo 8/73 (11.0%) 10/78 (12.8%) Total 40/340 (11.8%) 56/345 (16.2%) 9.2.9 Thus the MTCT rate increased by 4.4% in the non- breastfeeding group. And there were 16 confirmed cases of transmission (or death). It is possible that some of these cases were deaths rather than transmission and also that some women who claimed not to have breastfed, actually breastfed. It is also possible that the PCR test used at 6 weeks gave some false positives. A study by Glenda Gray on the influence of breastfeeding on MTCT found a number of indeterminate PCR results. A paper by Zijenah et. al., states that the use of PCR for diagnosis of HIV infection has been hampered by a lack of suitable primers for clade C viruses. In addition in September 2000 (after the Petra and SAINT PCR tests were done) Roche announced the development of an improved PCR kit. Thus virologists should be consulted to comment on the reliability of PCR particularly at 6-8 weeks. Any problems with this would weaken the interpretation of week 6-8 results compared to the month 18 or 20 results; in any case the latter are more useful in an operational / public health setting where one is interested in the overall long term effect of any MTCT strategy. _______________________________ AIDS Law Project c/o Legal Resources Centre 5th Floor, Centenery Building Bureau Lane PRETORIA Tel: (011) 717-8637 Fax: (011) 403-2341 Ref: A Kleinsmidt TO: The Registrar of the High Court PRETORIA AND TO: Attorney for 1st to 9th Respondents STATE ATTORNEY 4th Floor, Fedsure Forum Building South Tower 268 Van der Walt Street PRETORIA Ref: Ms G Behardien AND TO: Attorney for Tenth Respondent DE KLERK & VAN GEND CAPE TOWN c/o TIM DU TOIT & COMPANY INC., 19th Floor, Volkskas Building 230 Van der Walt Street PRETORIA Tel: (012) 320-6753 Fax: (012) 320-6746 Ref: Mr Botha/hs AND TO: Attorneys for Applicants G M BUDLENDER Legal Resources Centre JOHANNESBURG c/o LEGAL RESOURCES CENTRE Centenery Building Bureau Lane PRETORIA Tel: (011) 836-9831 Fax: (011) 834-4273 12 1