We appreciate the interest shown by Patel and colleagues in our recently published meta-analysis , in which we found no overall association between the use of antiretroviral therapy (ART) during pregnancy and premature delivery. Although a significantly increased risk of premature delivery was found with protease inhibitor (PI)-containing therapy compared with other combinations [odds ratio (OR) 1.35; 95% confidence interval (CI) 1.08–1.70] and with combination therapy initiated before or in the first trimester of pregnancy compared with therapy initiated later (OR 1.71; 95% CI 1.09–2.67), the magnitude of these effects was relatively modest.
We agree with Patel et al., and indeed mention in the discussion of our findings, that the meta-analysis needs to be interpreted in the context of the observational studies included and their methodological limitations. The limitations, as well as the heterogeneity among the studies, are explored in detail in the results and discussion . The statement that ‘no association’ was found ‘between prematurity and combination ART when duration of exposure was ignored’ does not accurately reflect the analysis that was performed. Our analysis included all eligible studies, not only those that did not include information on the duration of therapy. Only four studies included information on the time of ART initiation per trimester of pregnancy; these results are presented. The inclusion of different populations in a meta-analysis does not necessarily introduce bias; instead, it indicates a heterogeneity of the data addressing the question, which needs to be explored and understood for the interpretation and generalizability of the results. In our opinion, the most concerning potential cause of bias is that women on PI-containing combination therapy, particularly in earlier time periods, were likely to have more advanced HIV disease and to need potent combination therapy to treat their own infection. As advanced HIV disease is in itself a well-documented risk factor for premature delivery, it is difficult to know whether the risk noted is caused by confounding by the indication for treatment of maternal disease.
Patel and colleagues mention two recent studies (subsequent to our meta-analysis) from Europe that found an association of HAART and premature delivery. One of them is not yet published, and we look forward to seeing the results when they appear. The other  is a letter to the editor in response to the article by Cotter et al. , but unfortunately the number of preterm deliveries per ART category is not presented in the letter. A more recent study in the United States used data from the Pediatric Spectrum of HIV Disease , one of the largest pediatric cohorts. The authors report a decrease in preterm births from 35 to 22% in 1989 and 2004, respectively, among 11 231 mother–infant pairs, at a time when the use of ART in pregnancy increased from less than 2 to 82%. Of interest is the fact that compared with two-drug use, both no ART during pregnancy and three-drug ART that included a PI were found to be statistically significantly associated with an increased risk of preterm birth, in adjusted analysis, albeit with modest adjusted odds ratios of 1.16 (95% CI 1.02–1.32) for no ART and 1.21 (95% CI 1.04–1.40) for PI-triple ART. The addition of that study to our meta-analysis does not substantially change the direction or magnitude of the effect on the risk of premature delivery for any of the comparisons, with the exceptions of monotherapy compared with no ART, in which the protective effect becomes statistically significant (OR 0.83; 95% CI 0.74–0.93), and PI-containing versus other combinations, in which the harmful effect becomes statistically non-significant (OR 1.23; 95% CI 0.98–1.53). As no information was provided on the time of initiation of ART in pregnancy, this comparison could not be updated.
We agree with Patel et al. that their hypothesis of ART-induced cytokine shift from T helper type 2 to T helper type 1 cells causing premature delivery  is interesting, albeit very preliminary. Advanced HIV disease is characterized by a T helper type 2 cell predominance, but advanced HIV in itself has also been associated with increased rates of preterm delivery in several studies [4,6]. Furthermore, non-PI combination regimens can also successfully produce virological and immunological responses, thus raising a question as to why an association with premature delivery would be observed only with PI regimens. A careful study of a comprehensive cytokine panel among women with HIV infection who are on ART regimens for their own health or for prophylaxis should be carried out with the appropriate respective controls, and should include HIV-uninfected pregnant women.
As others before us have argued , randomized clinical trials of various ART regimens in pregnancy in women who do not need it for their own health would help to answer the question. In the likely absence of such a trial, an individual patient meta-analysis of existing evidence would be a welcome alternative. There is a need for greater collaboration among researchers in this field and for standardization of the analyses and presentation of results, including detailed information on risk per type of ART used and duration of use during pregnancy. For the time being and on the basis of all the available evidence, the current meta-analysis provides reassurance that there is no marked effect of ART therapy overall on increasing the risk of prematurity. PI-containing combinations and use early in pregnancy might, however, be associated with some increase in risk. Continued surveillance will be necessary in order to determine such a risk accurately.
Disclaimer: The findings and conclusions in this letters are those of the authors and do not necessarily reflect the views of the Centers for Disease Control and Prevention.
1. Kourtis AP, Schmid C, Jamieson DJ, Lau J. Use of antiretroviral therapy in pregnant HIV-infected women and the risk of premature delivery: a meta-analysis. AIDS 2007; 21:607–615.
2. Ravizza M, Martinelli P, Bucceri A, Fiore S, Alberico S, Tamburini E, et al
. Treatment with protease inhibitors and coinfection with hepatitis C virus are independent predictors of preterm delivery in HIV-infected pregnant women. J Infect Dis 2007; 195:913–914.
3. Cotter AM, Garcia AG, Duthely ML, Luke B, O'Sullivan MJ. Is antiretroviral therapy during pregnancy associated with an increased risk of preterm delivery, low birth weight, or stillbirth? J Infect Dis 2006; 193:1195–1201.
4. Schulte J, Dominguez K, Sukalac T, Bohannon B, Fowler MG. Declines in low birth weight and preterm birth among infants who were born to HIV-infected women during an era of increased use of maternal antiretroviral drugs: Pediatric Spectrum of HIV Disease, 1989–2004. Pediatrics 2007; 119:e900–e906.
5. Fiore S, Newell ML, Trabattoni D, Thorne C, Gray L, Savasi V, et al
. Antiretroviral therapy-associated modulation of Th1 and Th2 immune responses in HIV-infected pregnant women. J Reprod Immunol 2006; 70:143–150.
6. Stratton P, Tuomala RE, Abboud R, Rodriguez E, Rich K, Pitt J, et al
. Obstetric and newborn outcomes in a cohort of HIV-infected pregnant women: a report of the Women and Infants Transmission Study. J Acquir Immune Defic Syndr Hum Retrovirol 1999; 20:179–186.
7. Tuomala RE, Yawetz S. Protease inhibitor use during pregnancy: is there an obstetrical risk? J Infect Dis 2006; 193:1191–1194.