Effects of antituberculosis treatment on pregnancy outcomes in infertile women with genital tuberculosis: a systematic review

Inclusion criteria

We included randomised controlled trials (RCTs) and expanded the inclusion criteria to non-RCTs, and comparative cohort studies in any language. The decision was made because only a small number of randomised trials could be available as found in the preliminary search that we conducted. We included studies that compared antituberculosis treatment against not receiving antituberculosis treatment or receiving the treatment for a shorter period in infertile women with genital tuberculosis without active tuberculosis disease and reported data on pregnancy outcomes. The antituberculosis treatment of interest was the scheme based on, isoniazid, rifampicin, pyrazinamide and ethambutol (HRZE) for 2 months, followed by isoniazid and rifampicin (HR) for the next 4 months. As detailed in the protocol, we assess pregnancy outcomes including pregnancy, full-term pregnancy, ectopic pregnancy and abortion or intrauterine death, as well as adverse events. We excluded case–control or cross-sectional studies, case reports, reviews, editorials or letters that do not present original results, and articles not available in full text.

Literature search and study selection

We searched in PubMed/MEDLINE, CENTRAL and EMBASE until 15 January 2023, with no restriction on language or publication date. The full search strategy for each database is available in online supplemental file 1. The elimination of duplicates and the selection of studies were performed using online software (Rayyan QCRI, Qatar Computing Research Institute). Two authors (DF-G and SAM-R) independently screened titles and abstracts to identify potentially relevant studies for inclusion. These potential studies were full text reviewed independently by two authors (BC-C and AGG-U). The disagreements were resolved with a third author (KFL). Excluded studies and reasons for exclusion can be found in online supplemental file 2. In addition, we manually search the reference lists of included studies to identify more relevant studies.

Data extraction

Two authors (KFL and KGT) independently extracted the following information from each included study using a previously designed Microsoft Excel sheet: first author, year of publication, country, study design, study period and follow-up, diagnosis of tuberculosis, sample size, age, intervention and comparator details, funding, and outcomes (pregnancy, at-term pregnancy, ectopic pregnancy, abortion or intrauterine death and adverse events). Disagreements were resolved with a third author (DRS-M).

Risk of bias

Two authors (DRS-M and JF-M) independently assessed the risk of bias in the included studies using the Cochrane Risk of Bias (RoB) 1.0 tool for RCT,20 and the Risk Of Bias In Non-randomised Studies (ROBINS-I) for non-RCT.21 Disagreements were resolved with a third author (SGL). The Cochrane RoB 1.0 tool consists of seven items: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessors, incomplete outcome data, selective reporting of results, and other sources of bias (unbalance in basal characteristics). For each item, the decision may be of low, high or unclear risk of bias.20 The ROBINS-1 consists of seven items. For each item, the decision may be low risk of bias, moderate risk of bias, serious risk of bias, critical risk of bias and no information.21 We did not use the New-Castle Ottawa Scale tool because no comparative cohort studies were included.

Statistical analysis

We used Stata V.16.0 to calculate risk ratios (RRs) with 95% CIs and Grading of Recommendations, Assessment, Development and Evaluations (GRADE) pro Software to calculate absolute effects with 95% CI for each outcome. We could not perform a prespecified meta-analysis due to the number of included studies and heterogeneity between them in terms of study design, type of comparator and population (women with genital tuberculosis without and with structural damage). In consequence, we could not perform the heterogeneity and publication bias statistical assessment.

We could not perform prespecified subgroup analysis according to country, mode of diagnosis of genital tuberculosis and history of previous tuberculosis treatment due to lack of data. We presented the results separately for women with structural damage, defined as the presence of lesions characteristic of tuberculosis in the ovary, tubes or uterus4; and for those without structural damage, defined as having only a positive endometrial PCR test.22

Certainty of evidence

We used the GRADE approach to assess the certainty of the evidence for each outcome as high, moderate, low or very low, based on study limitations, imprecision, risk of bias, inconsistency, indirectness and publication bias.23 With GRADE, RCT and non-RCT designs begin at as high and low certainty of the evidence, respectively.24 Then, we used the following criteria to rating down the certainty of evidence: regarding imprecision, we used the sample size rather than setting thresholds in a minimally contextualised approach.25 For this purpose, we followed the criteria proposed by Murad et al to rating the certainty of evidence in the absence of a single estimate of effect.26 We rated down one level for a dichotomous outcome if the study had less than 300 events and two levels if the study had less than 50 events. Regarding the risk of bias, we rated down one level if the study had fewer than six items with a low risk of bias and by two levels if the study had fewer than five items with a low risk of bias. The inconsistency and publication bias could not be assessed because one study was included for each outcome. We presented the results in Summary of Findings tables and followed the GRADE guidance for communicating the results.27

Patient and public involvement

Patients and public were not involved in this study.

Selection of studies

We found 1468 articles in the systematic search. After the removal of duplicates, we examined 970 records by title and abstract, of which 20 records were reviewed in full text. Finally, we included three studies.22 28 29 We did not find additional studies after searching the references of the included studies (figure 1).

Figure 1

Flow diagram summarising the process of literature search and selection.

Characteristics of the studies

Two studies were RCT,22 29 and one was non-RCT.28 The sample size ranged from 78 to 175 participants. One study22 evaluated infertile women without structural damage with positive endometrial PCR test for tuberculosis and two studies28 29 evaluated infertile women with structural damage according to the hysteroscopy or laparoscopy findings. The mean age ranged from 28 to 29 years. Two studies22 28 evaluated the effectiveness of antituberculosis therapy with HRZE for a period of 6–12 months compared with no antituberculosis treatment, followed by assisted reproductive therapies. One study29 evaluated different regimens of antituberculosis treatment with HRZE comparing the efficacy of treatment for 9 vs 6 months. The funding of the studies and other characteristics are listed in table 1.

Risk of bias

Overall, the studies were rated as unclear or high risk of bias. The RCT of Kriplani et al22 presented unclear risk of bias in allocation concealment and a high risk of bias in blinding participants, study personnel and outcome assessors. The RCT of Sharma et al29 presented an uncertain risk of bias in blinding of participants, study personnel and outcome assessors, and selective reporting of results (figure 2).

Figure 2

Risk of bias.

The non-RCT of Yue et al28 had critical risk in the domain of bias due to confounding and bias due to deviations from the intended interventions. In addition, the study had serious risk in the domains of bias due to participant selection and bias in the classification of interventions. There was no information on bias due to missing data. The risk was moderate in the domain bias in the measurement of outcomes and there was low risk in the bias domain assessing reported outcome selection (figure 2).

Antituberculosis treatment for women with genital tuberculosis without structural damage

Very low certainty of evidence from one RCT (100 participants)22 shows that at 12 months follow-up, the antituberculosis treatment (HRZE) for 6 months, compared with no antituberculosis treatment, may have little to no effect on pregnancy (71 more per 1000, 95% CI −88 to +311), full-term pregnancy (34 more per 1000, 95% CI −103 to +232), abortion or intrauterine death (11 more per 1000, 95% CI −3 to +36) and ectopic pregnancy (not estimable) but the evidence is very uncertain. In addition, the study reported no adverse events in either group (table 2).

Antituberculosis treatment for women with genital tuberculosis with structural damage

Very low certainty of the evidence from one non-RCT (78 participants)28 shows that at no detailed follow-up the antituberculosis treatment (HRZE) for 6–12 months, compared with no antituberculosis treatment, may reduce the pregnancy rate (297 fewer per 1000, 95% CI −416 to −101) but the evidence is very uncertain. In addition, the study did not report information about adverse events (table 3).

Other comparisons: 9-month vs 6-month antituberculosis treatment for women with genital tuberculosis

Very low certainty of evidence from one RCT (175 participants)29 shows that the antituberculosis treatment for 9 months (HRZE), in comparison to antituberculosis treatment for 6 months, have little to no effect on pregnancy (13 more per 1000, 95% CI −56 to +107), at term pregnancy (9 more per 1000, 95% CI −35 to +75), abortion or intrauterine death (1 more per 1000, 95% CI−26 to +52), and preterm pregnancy (1 more per 1000, 95% CI −22 to +48) but the evidence is very uncertain. In addition, the study reported that adverse events were similar in both groups (online supplemental file 3.

Summary of main results

We included one RCT and one non-RCT that compared the effectiveness of antituberculosis treatment compared with no antituberculosis treatment in infertile women with genital tuberculosis without and with structural damage, respectively. In infertile women without structural damage, the antituberculosis treatment for 6 months may have little to no effect on pregnancy outcomes and adverse events, but the evidence is very uncertain. In contrast, in infertile women with structural damage, the antituberculosis treatment for 6–12 months may reduce the pregnancy rate, but the evidence is very uncertain. In addition, we included one RCT that compared a 9-month vs 6-month antituberculosis treatment regimen was included. However, the evidence is very uncertain.

Effects in infertile women without structural damage

The use of antituberculosis treatment to improve pregnancy outcomes was proposed based on biological plausibility. However, the included RCT found that the effects of the antituberculosis treatment given for 6 months on pregnancy outcomes are very uncertain.22 This could be explained because most potentially useful drugs based on biological plausibility ultimately do not cause clinical benefits in patients. However, all can produce adverse events. In this sense, a study that evaluated articles published in six major basic science journals between 1979 and 1983 found that of 101 novel or promising therapies, 27 resulted in the publication of at least one clinical trial, of which 5 were approved for use, and of which only 1 (0.99% of promising therapies) shown clear clinical benefits.30

Effects in infertile women with structural damage

The included non-RCT showed that, with very low certainty of evidence, antituberculosis treatment for 6–12 months may reduce the pregnancy rate. This effect could be explained by possible imbalances in the characteristics of the participants and the small sample size of the study. Although no statistically significant differences were found in the characteristics of the participants, women who received antituberculous treatment, compared with those who did not, tended to have a higher frequency of hydrosalpinx (41.3% (19/46) vs 40.6% (13/32)), pelvic adhesions (97.8% (45/46) vs 96.9% (31/32)) and age greater than or equal to 30 years (30.4% (14/46) vs 28.1% (9/32)), respectively.28 Based on this, the probability of pregnancy could be lower in women who received antituberculosis treatment. It has been described that small studies, without randomisation, and other methodological limitations tend to find statistically significant and biased results.31 Therefore, these results should be interpreted with caution.

On the other hand, antituberculosis treatment may not have a significant impact in reducing severe lesions such as adhesions, which would be the main mechanical causes of infertility. In this sense, a Canadian cohort that included infertile women with genital tuberculosis found that the frequency of tubercles in the pelvic peritoneum, fallopian tube and ovary was reduced from 54% to 2.04% after antituberculosis treatment. However, the frequency of pelvic adhesions was similar before and after treatment (42% vs 42.5%, respectively).10

Implications

Currently, the treatment of choice to achieve pregnancy in women without active infection is assisted fertilisation techniques. For women without structural damage, less invasive methods such as ovulation induction and timed intercourse are proposed.6 On the other hand, in women with structural damage to a functional endometrium, in vitro fertilisation and embryo transfer could be the best therapeutic option.30 If the endometrium is damaged but the ovaries are intact, gestational surrogacy could be the treatment of choice.4 13 However, these techniques have a pregnancy rate of 60.6%.16 Given this, the studies proposed using antituberculosis treatment with HRZE before assisted fertilisation techniques to improve pregnancy rates.

With a very low certainty of evidence, the results of this systematic review showed that the effects of antituberculosis treatment on pregnancy outcomes are very uncertain. In contrast, although no frequent adverse events were reported, it is important to note that the adverse events could be underestimated due to small sample size in the studies. In this sense, a multicentric cohort study in Morocco that included 2532 participants, found that the incidence of adverse events of antituberculosis treatment was 10% at 6 months follow-up.32 In addition, the impact of the treatment’s cost could be variable between the different health systems worldwide. A review published in 2015 found the cost of the antituberculosis treatment range from US$14 659 in high-income countries, US$840 in upper-middle-income countries, US$273 in lower-middle-income countries and US$258 in low-income countries.33

Finally, the use of antituberculous treatment could delay of 6–12 months in initiating assisted fertility treatments. Therefore, it is likely that the delay in starting assisted fertility techniques could reduce the probability of pregnancy. This would be especially important, as time is a key factor in achieving pregnancy, especially in older women.22

Limitations of the included studies

The body of evidence has limitations. The studies were heterogeneous in terms of population, intervention, control groups and outcomes assessed, making comparison difficult.

All the studies were probably underpowered to evaluate adverse events and the reporting of the adverse events was poor. Overall, the certainty of the evidence was very low for all results mainly because of the risk of bias and imprecision. In addition, two studies were conducted in India and one in China, countries with a high burden of tuberculosis,34 and all the studies had restrictive inclusion criteria. Therefore, the results should be interpreted with caution when extrapolating to other groups.

Recommendations for future studies

Future RCT should be double-blinded and include a larger number of participants to avoid concerns about the risk of bias and imprecision. The effect of the treatment should be evaluated according to whether the patient has structural damage. Finally, the studies should report the type of pregnancy and adverse events.

Limitations and strengths in the review process

This systematic review has limitations. We did not search regional databases from tuberculosis-endemic countries, so some studies may have been missed. Nevertheless, we searched in three main databases without language restriction, and we extended the search screening references of included studies to find other relevant trials. The evidence is limited and comes from only two RCTs and one non-RCT with methodological limitations, resulting in a very low certainty of evidence in their results. Not all included studies provided detailed information about their methodology, evaluated the effects for all pregnancy outcomes, or reported the adverse events of the therapy. Finally, a meta-analysis could not be performed due to the small number of studies and heterogeneity among them and publication bias was not statistically assessed since less than 10 trials were included.35

Strengths of this review include that, to our best knowledge, this is the first systematic review that assesses the effects of antituberculosis on clinically relevant pregnancy outcomes for infertile women with genital tuberculosis without active tuberculosis disease and clinicians to inform the decision-making process. Also, we used rigorous methods according to the Cochrane Handbook for Systematic Reviews of Intervention, including the GRADE approach to assess the certainty of the evidence.35