Background
Spontaneous preterm birth remains a leading cause of neonatal morbidity and mortality among nonanomalous neonates in the United States. Spontaneous preterm birth tends to recur at similar gestational ages. Intramuscular 17-alpha hydroxyprogesterone caproate reduces the risk of recurrent spontaneous preterm birth. Unfortunately, one-third of high-risk women will have a recurrent spontaneous preterm birth despite 17-alpha hydroxyprogesterone caproate therapy; the reasons for this variability in response are unknown.
Objective
We hypothesized that clinical factors among women treated with 17-alpha hydroxyprogesterone caproate who suffer recurrent spontaneous preterm birth at a similar gestational age differ from women who deliver later, and that these associations could be used to generate a clinical scoring system to predict 17-alpha hydroxyprogesterone caproate response.
Study Design
Secondary analysis of a prospective, multicenter, randomized controlled trial enrolling women with ≥1 previous singleton spontaneous preterm birth <37 weeks’ gestation. Participants received daily omega-3 supplementation or placebo for the prevention of recurrent preterm birth; all were provided 17-alpha hydroxyprogesterone caproate. Women were classified as a 17-alpha hydroxyprogesterone caproate responder or nonresponder by calculating the difference in delivery gestational age between the 17-alpha hydroxyprogesterone caproate–treated pregnancy and her earliest previous spontaneous preterm birth. Responders were women with pregnancy extending ≥3 weeks later compared with the delivery gestational age of their earliest previous preterm birth; nonresponders delivered earlier or within 3 weeks of the gestational age of their earliest previous preterm birth. A risk score for nonresponse to 17-alpha hydroxyprogesterone caproate was generated from regression models via the use of clinical predictors and was validated in an independent population. Data were analyzed with multivariable logistic regression.
Results
A total of 754 women met inclusion criteria; 159 (21%) were nonresponders. Responders delivered later on average (37.7±2.5 weeks) than nonresponders (31.5±5.3 weeks), P <.001. Among responders, 27% had a recurrent spontaneous preterm birth (vs 100% of nonresponders). Demographic characteristics were similar between responders and nonresponders. In a multivariable logistic regression model, independent risk factors for nonresponse to 17-alpha hydroxyprogesterone caproate were each additional week of gestation of the earliest previous preterm birth (odds ratio, 1.23; 95% confidence interval, 1.17−1.30, P <.001), placental abruption or significant vaginal bleeding (odds ratio, 5.60; 95% confidence interval, 2.46−12.71, P <.001), gonorrhea and/or chlamydia in the current pregnancy (odds ratio, 3.59; 95% confidence interval, 1.36−9.48, P =.010), carriage of a male fetus (odds ratio, 1.51; 95% confidence interval, 1.02−2.24, P =.040), and a penultimate preterm birth (odds ratio, 2.10; 95% confidence interval, 1.03−4.25, P =.041). These clinical factors were used to generate a risk score for nonresponse to 17-alpha hydroxyprogesterone caproate as follows: black +1, male fetus +1, penultimate preterm birth +2, gonorrhea/chlamydia +4, placental abruption +5, earliest previous preterm birth was 32−36 weeks +5. A total risk score >6 was 78% sensitive and 60% specific for predicting nonresponse to 17-alpha hydroxyprogesterone caproate (area under the curve=0.69). This scoring system was validated in an independent population of 287 women; in the validation set, a total risk score >6 performed similarly with a 65% sensitivity, 67% specificity and area under the curve of 0.66.
Conclusions
Several clinical characteristics define women at risk for recurrent preterm birth at a similar gestational age despite 17-alpha hydroxyprogesterone caproate therapy and can be used to generate a clinical risk predictor score. These data should be refined and confirmed in other cohorts, and women at high risk for nonresponse should be targets for novel therapeutic intervention studies.
The proportion of babies born preterm in the United States remains unacceptably high at 11.4% (2013). The majority of preterm deliveries are spontaneous preterm birth (SPTB); this encompasses deliveries caused by preterm premature rupture of membranes, cervical insufficiency, and idiopathic preterm labor. A history of a previous SPTB is the greatest risk factor for preterm birth; SPTB recurs in 35−50% of women and tends to recur at similar gestational ages. The probability of SPTB also increases with the number of previous SPTBs a woman has experienced, with the most recent birth being the most predictive.
In 2003, Meis et al published results from a multicenter, randomized controlled trial of intramuscular 17-alpha hydroxyprogesterone caproate (17-OHPC), demonstrating that weekly intramuscular injections beginning at 16−20 weeks’ gestation reduces the risk of recurrent prematurity by approximately one-third. Unfortunately, prophylactic 17-OHPC is not effective for all women, because one-third of high-risk women will have a recurrent episode of preterm birth (PTB) despite appropriate prophylaxis with 17-OHPC; the reasons for this variable response are unknown. Limited genetic studies have found associations between genetic variation in the progesterone receptor and nitric oxide genes and clinical response to 17-OHPC. Other studies have focused on clinical factors in attempts to define the population of women most likely to respond to treatment, with those authors finding that women with a family history of PTB and those who experienced bleeding or abruption in the current pregnancy were less likely to respond to 17-OHPC and those who had a previous spontaneous PTB <34 weeks were more likely to deliver at term with 17-OHPC.
We sought to compare demographic, historical, and antenatal factors between women who delivered at similar gestational age with 17-OHPC for recurrent SPTB prevention vs those women who deliver later with 17-OHPC.
Materials and Methods
This is a secondary analysis of a multicenter randomized controlled trial of daily omega-3 supplementation vs placebo for recurrent preterm birth prevention conducted by the Eunice Kennedy Shriver National Institute for Child Health and Human Development Maternal Fetal Medicine Units Network. To summarize, women pregnant with a singleton, nonanomalous fetus who had a history of a previous documented singleton SPTB between 20 weeks 0 days’ gestation and 36 weeks 6 days’ gestation were recruited between 16 and 22 weeks’ gestation across 13 clinical sites between January 2005 and October 2006. All participants received weekly 17-OHPC (beginning at randomization) as a part of the study and were randomized to receive either a daily supplement of 1200 mg of eicosapentaenoic acid (20:5n-3) and 800 mg of docosahexaenoic acid (22:6n-3), for a total of 2000 mg of omega-3 long chain polyunsaturated fatty acids (omega-3 group), divided into 4 capsules, or matching placebo capsules.
The 17-OHPC injections were supplied by a company that manages investigational drugs (Eminent Services, Frederick, MD) and were continued until delivery or 36 weeks 6/7 days of gestation, whichever occurred first. Compliance with 17-OHPC injections was calculated as the number received divided by the expected number. Obstetric management was otherwise performed per each woman’s primary obstetric provider. The main study did not find any impact on the rate of PTB before 37, 35, or 32 weeks with omega-3 supplementation among women already receiving 17-OHPC. Comprehensive obstetric history, longitudinal antenatal data, and detailed delivery information were collected at the time the main study was conducted.
Participants in the original study had the option of providing a maternal peripheral blood DNA sample for a planned analysis of the relationship between maternal genotype at the –308 position of the tumor necrosis factor-α gene, the –174 position of the interleukin (IL)-6 gene, and the +3954 position of the IL-1β gene and length of gestation and risk of extreme preterm delivery. The procedure for DNA extraction and sample genotyping was performed in accordance with standard methodology. The authors found that women homozygous for the minor allele (A) at the –308 position in the promoter region of the tumor necrosis factor-gene had a significantly shorter length of gestation.
This secondary analysis used a deidentified data set, and after review by the Institutional Review Board at the University of North Carolina, was considered exempt from institutional research board oversight.
For this analysis, we included all women who had a documented earliest gestational age of previous PTB and with gestational age outcome information available in the current pregnancy. Gestational age was determined by a combination of last menstrual period (if available) and ultrasound. We excluded women who received fewer than 50% of expected 17-OHPC injections, regardless of delivery gestational age.
We classified women as a 17-OHPC responder or nonresponder by calculating the difference in the gestational age at delivery between the 17-OHPC−treated pregnancy and her earliest SPTB, as we have described previously. To review, the difference between the earliest delivery gestational age and the delivery gestational age with 17-OHPC was calculated and termed the “17-OHPC effect.” Women with a 17-OHPC effect of ≥3 weeks (ie, the individual’s pregnancy or pregnancies treated with 17-OHPC delivered at least 3 weeks later compared with the gestational age of the earliest PTB without 17-OHPC treatment) were considered 17-OHPC responders; this designation was made in part with the use of data originally published by Bloom et al, which showed that 70% of women with recurrent PTB will experience recurrence within 2 weeks of their initial PTB. Women with a negative overall 17-OHPC effect and those with an overall 17-OHPC effect of <3 weeks were classified as nonresponders. Women with a 17-OHPC effect of <3 weeks but delivering at term (for example, if the earliest previous SPTB was at 35 weeks, and the patient delivered at 37 weeks during the 17-OHPC−treated pregnancy) were considered “equivocal” 17-OHPC responders and were excluded from analysis. A separate limited subgroup analysis that examined the subset of nonresponders who delivered significantly earlier (≤3 weeks earlier) during the 17-OHPC−treated pregnancy also was performed (“possible harm”) group.
Bivariate analyses were conducted to compare 17-OHPC responders with nonresponders with the t test, χ 2 , or Fisher exact test, as appropriate. Genotype data were assessed with the additive model of inheritance (each copy of the minor allele was hypothesized to confer additional risk), and in the bivariate analysis, were analyzed separately by self-reported race/ethnicity group to reduce the chance of analytic error due to population stratification. Multivariable logistic regression analysis was performed to identify factors associated with nonresponse to 17-OHPC, by the use of backwards variable selection, and we retained all factors with P <.20 in the final model.
Finally, a clinical prognostic score was developed. We used the final multivariable model as described in the methods previously and then retained only factors readily available during the antenatal period (eg, we excluded genotype information and any data only available postnatally). A score was then derived for each prognostic variable with odds ratio (OR) estimates; each OR coefficient was divided by the smallest OR coefficient and rounded up to the nearest 0.5 to produce a weighted score. Each woman was then assigned a risk score with this calculation, and the value of this score in predicting nonresponse to 17P was then assessed. A statistical “cut-point” was determined based on the risk score where the sensitivity and specificity were maximized, via use of the Liu method. The model was then validated with deidentified data from the randomized trial of 17-OHPC vs placebo conducted by the Eunice Kennedy Shriver National Institute for Child Health and Human Development Maternal Fetal Medicine Units Network.
All statistical analyses were performed with STATA version 13.1 (College Station, TX). For modeling, fewer than 5% of data were missing, so missing data were imputed by replacing missing values with the median for continuous variables and with the most frequent category for categorical variables, which works as well as multiple imputation with this small proportion of missing data. A P value of <.05 was considered statistically significant.
Results
From the original cohort of 852 women enrolled in the omega-3 randomized controlled trial (RCT), 754 met inclusion criteria ( Figure ). Of note, 57 of 852 (6.7%) women from the original cohort had a 17-OHPC effect <3 weeks yet delivered at term because their earliest previous PTB was more than 34 weeks’ gestation; these women were considered “equivocal” responders and were excluded ( Figure ). Of the remaining women, 595 of 754 (78.9%) delivered at least 3 weeks later compared with the delivery gestational age of their previous earliest PTB and were considered 17-OHPC responders, and 159 (21.1%) had a 17-OHPC effect <3 and were considered nonresponders. Of the 159 nonresponders, 31 (19.4%) delivered ≥3 weeks earlier with 17-OHPC. On average, responders delivered later (37.7±2.5 weeks) than nonresponders (31.5±5.3 weeks), P <.001. Overall, 160 of 595 responders (26.9%) had a recurrent SPTB, but they were considered responders because they delivered ≥3 weeks later compared with their earliest SPTB.
Many demographic characteristics were similar between responders and nonresponders, although nonresponders were more likely to be white and less likely to be of Hispanic ethnicity ( Table 1 ). The gestational age of the earliest previous SPTB was later among nonresponders (32.3±4.1 vs 29.1±4.5, P <.001), but both groups had a similar number of previous SPTB (1.4±0.6 vs 1.4±0.7, P =.597), similar chance of having had multiple SPTB (34.6% vs 28.4%, P =.129), and similar proportion had 1 or more previous term deliveries (28.9% vs 35.3%, P =.132). The majority of women had a penultimate pregnancy delivering preterm; this did not vary by responder status (92.3% vs 87.8%, P =.121, Table 1 ).
| 17-OHPC nonresponder (n=159) | 17-OHPC responder (n=595) | P value | |
|---|---|---|---|
| Maternal age, mean, ± SD | 27.1±5.8 | 27.9±5.5 | .121 |
| Maternal race, n (%) | .001 | ||
| Black | 53 (33.3) | 199 (33.5) | |
| White | 93 (58.5) | 278 (46.7) | |
| Other | 13 (8.2) | 118 (19.8) | |
| Maternal Hispanic ethnicity, n (%) | 11 (6.9) | 100 (16.8) | .002 |
| Gestational age of earliest previous spontaneous preterm birth, mean ± SD | 32.3±4.1 | 29.1±4.5 | <.001 |
| Penultimate pregnancy delivered preterm, n (%) | 147 (92.5) | 525 (88.2) | .121 |
| Number of previous spontaneous preterm births, mean ± SD | 1.4±0.6 | 1.4±0.7 | .597 |
| More than 1 previous spontaneous preterm birth, n (%) | 55 (34.6) | 169 (28.4) | .129 |
| One or more previous term birth(s), n (%) | 46 (28.9) | 210 (35.3) | .132 |
| Gestational age at randomization and initiation of 17-OHPC, wk, mean ± SD | 19.5±1.7 | 19.3±1.7 | .198 |
| 17-OHPC injection compliance, median % (IQR) | 100 (92, 100) | 100 (93, 100) | .022 |
Several current pregnancy factors also were similar between 17-OHPC responders and nonresponders; however, nonresponders were more likely to experience clinically significant vaginal bleeding or be diagnosed with a placental abruption in the current gestation (11.3% vs 3.2%, P <.001) and also were more likely to be diagnosed with gonorrhea and/or chlamydia during pregnancy (6.9% vs 2.2%, P =.003, Table 2 ). As expected, because of the greater rate of PTB among nonresponders, this group also was more likely to receive antenatal corticosteroids (27.0% vs 28.5%, P =.017), tocolysis (32.7% vs 18.0%, P <.001), and be admitted to the hospital for tocolysis or preterm labor symptoms ( Table 2 ). Nonresponders were also more likely to carry a male fetus (57.9% vs 47.9%, P =.026, Table 2 ).
| 17-OHPC nonresponder (n=159) | 17-OHPC responder (n=595) | P value | |
|---|---|---|---|
| Prepregnancy body mass index, kg/m 2 , mean ± SD | 26.0±6.6 | 26.7±6.7 | .284 |
| Smoked during pregnancy, n (%) | 26 (16.4) | 94 (15.8) | .865 |
| One or more episode(s) of vaginal bleeding or diagnosis of clinical abruption, n (%) | 18 (11.3) | 19 (3.2) | <.001 |
| Bacterial vaginosis, n (%) | 11 (23.4) | 43 (33.1) | .217 |
| Trichomonas during pregnancy, n (%) | 2 (4.3) | 14 (10.8) | .182 |
| Gonorrhea and/or chlamydia infection during pregnancy, n (%) | 11 (6.9) | 13 (2.2) | .003 |
| Short cervix <2.50 cm, n (%) a | 4 (6.8) | 9 (3.0) | .243 |
| Cervical cerclage, n (%) | 4 (2.5) | 12 (2.0) | .698 |
| Randomized to receive omega-3 capsules, n (%) | 81 (50.9) | 301 (50.6) | .937 |
| Received antenatal corticosteroids, n (%) | 43 (27.0) | 110 (18.5) | .017 |
| Received tocolysis, n (%) | 52 (32.7) | 107 (18.0) | <.001 |
| Median number of hospital admissions for tocolysis (IQR) | 0 (0, 1) | 0 (0, 0) | <.001 |
| Male fetus | 92 (57.9) | 285 (47.9) | .026 |
a Among 355 women with at least 1 transvaginal cervical length assessment.
Next, we evaluated the relationship among the 3 maternal genotypes and 17-OHPC response ( Tables 3 and 4 ). For both black women and white/Hispanic women, there was a trend towards a relationship between maternal IL-1β+3954 genotype and 17-OHPC response, but this did not reach statistical significance ( Tables 3 and 4 ).
| 17-OHPC nonresponders (n=50) | 17-OHPC responders (n=198) | P value | |
|---|---|---|---|
| IL-6-174 a , n (%) | .115 | ||
| CC a | 1 (2.0) | 0 (0.0) | |
| GC | 9 (18.0) | 44 (22.2) | |
| GG | 40 (80.0) | 154 (77.8) | |
| TNF-α-308 b , n (%) | .361 | ||
| AA b | 1 (2.0) | 4 (2.0) | |
| AG | 7 (14.0) | 46 (23.2) | |
| GG | 42 (84.0) | 148 (74.8) | |
| IL-1β+3954 b , n (%) | .062 | ||
| TT b | 3 (6.0) | 3 (1.5) | |
| CT | 7 (14.0) | 49 (24.8) | |
| CC | 40 (80.0) | 146 (73.7) |
a Genotype data available for 248 of 252 (98%) of black women
| 17-OHPC nonresponders (n=103) | 17-OHPC responders (n=363) | P value | |
|---|---|---|---|
| IL-6-174 b , n (%) | .990 | ||
| CC a | 14 (13.6) | 48 (13.1) | |
| GC | 47 (46.6) | 169 (46.2) | |
| GG | 42 (40.8) | 149 (40.7) | |
| TNF-α-308 b , n (%) | .157 | ||
| AA b | 4 (3.9) | 4 (1.1) | |
| AG | 27 (26.2) | 95 (26.2) | |
| GG | 72 (69.9) | 264 (72.7) | |
| IL-1β+3954 a , n (%) | .051 | ||
| TT b | 8 (7.8) | 16 (4.4) | |
| CT | 41 (39.8) | 112 (30.6) | |
| CC | 54 (52.4) | 238 (65.0) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
