Vasoactive exposures and risk of amniotic band syndrome and terminal transverse limb deficiencies

Abstract

Introduction:

Amniotic band syndrome (ABS) includes limb deficiencies accompanied by fibrous strands originating from the amniotic lining. Terminal transverse limb deficiencies (TTLD) appear to be similar but lack fibrous strands. Both are hypothesized to result from vascular disruption. For ABS, limb deficiencies are considered secondary to amnion rupture. We explored an alternative possibility—that TTLD is the primary defect and ABS is secondary.

Methods:

Using data from the National Birth Defects Prevention Study, we expanded on a previous study. We examined smoking, alcohol, and medications categorized by indicated vasoactivity as markers of vascular disruption. Logistic regression models with Firth's penalized likelihood were used to estimate adjusted odds ratios (aORs) and 95% confidence intervals (CIs).

Results:

Use of bronchodilators and aspirin appeared to increase the risk of ABS, while decongestants and nonaspirin NSAIDs increased the risk of TTLD. The risk of ABS was markedly increased in cases reporting combinations of vasoactive exposures, particularly alcohol and aspirin (aOR 3.7, 95% CI 1.6, 7.8), and alcohol and bronchodilators (aOR 3.4, 95% CI 1.4, 7.5). Increased risk of TTLD due to combinations of vasoactive exposures was only observed for smoking and decongestants (aOR 2.3, 95% CI 1.4, 3.6).

Conclusions:

Exposures associated with increased risk of ABS had no apparent association with TTLD, supporting previous evidence that these may be distinct phenotypes. ABS appears to be associated with combined exposures with vasodilation properties, such as alcohol and bronchodilators, while increased risk of TTLD may be associated with smoking and decongestants, both vasoconstrictive exposures.

1 ∣. INTRODUCTION

Amniotic band syndrome (ABS) includes limb malformations accompanied by the presence of fibrous strands of tissue originating from the amniotic lining (Holmes, Westgate, Nasri, & Toufaily, 2018; Van Allen, 1981). ABS is typically evidenced by a deformity of mainly the limbs and often includes the presence of constriction rings, some degree of syndactyly, and occasionally fibrous bands encircling one or more fingers (Holmes et al., 2018). Terminal transverse limb deficiencies (TTLD) are similar in appearance to ABS but lack fibrous strands, and are considered an “amputation” of an arm, leg, or finger (Holmes et al., 2018). Etiologies of ABS and TTLD are not well-understood and different pathogenic mechanisms have been proposed. Both ABS and TTLD are among several structural abnormalities attributed to a “vascular disruption” process during gestation (Garza, Cordero, & Mulinare, 1988; Husain, Langlois, Sever, & Gambello, 2008; Van Allen, 1981; Werler, Bosco, & Shapira, 2009; Werler, Louik, & Mitchell, 2003; Werler, Sheehan, Hayes, Mitchell, & Mulliken, 2004; Werler, Sheehan, & Mitchell, 2003), in which localized disruption of embryonic and fetal blood cells leads to structural anomalies and tissue deficiencies in a previously normal limb structure (Holmes, 2011; Holmes et al., 2018). While some structures eventually heal, others result in abnormalities such as constriction rings, distal tissue loss or syndactyly (Holmes, 2011). The timing and severity of the underlying disruptive mechanism dictate the nature of the resulting abnormalities (Holmes et al., 2018; Los, Brandenburg, & Niermeijer, 1999; Van Allen, 1992).

For ABS, amnion rupture has traditionally been considered the primary defect, which launches a cascade of vascular disruptive events and leads, in some cases, to TTLD as a secondary outcome (Torpin, 1965, 1968; Van Allen, 1981). Evidence in support of this are experimental studies of amnion rupture which induced TTLDs (Kennedy, 1977). In the 1990s, necrotic lesions in extremities resulting from cocaine-induced vasoconstriction were observed in rodent model (Webster & Brown-Woodman, 1990), prompting us to consider an alternative hypothesis: TTLD is the primary defect and associated necrotic lesions adhere to and tear the amnion, leading to fibrous bands (ABS) as a secondary outcome. It remains unclear whether the amniotic bands are a result of the vascular disruptive process or precede it (McGuirk, Westgate, & Holmes, 2001; Moerman, Fryns, Vandenberghe, & Lauweryns, 1992). Though both ABS and TTLD may have more than one etiology; if the prevailing mechanism is that the fibrous bands result from vascular disruption-induced limb necrosis, then TTLD and ABS could be considered as primary and secondary manifestation of the same underlying defect. If the epidemiologic profiles and risk patterns are different, then perhaps the pathogenesis of these limb deficiencies include one or more of the previously proposed mechanisms.

A previous study examined the epidemiologic characteristics of limb defects with and without evidence of amniotic bands to compare patterns between the case groups of maternal characteristics and vasoactive exposures, which served as possible markers for vascular disruption, using data on 1997–2004 births in the National Birth Defects Prevention Study (NBDPS) (Werler et al., 2009). The findings demonstrated that maternal demographic, clinical and vasoactive risk factors differed between ABS and TTLD cases, suggesting that these abnormalities may be distinct entities (Werler et al., 2009). Maternal vasoactive exposures were not uniformly associated with either case group but estimates were based on small numbers.

The current study extends the previous NBDPS analysis with additional years of data (2005–2011). In addition to cigarette smoking and vasoactive medications from the earlier analysis, we also included maternal alcohol consumption. Based on previously reported increased risks of other vascular birth defects due to the combination of vasoactive medication use and cigarette smoking (Werler et al., 2004; Werler, Louik, & Mitchell, 2003; Werler, Sheehan, & Mitchell, 2003), we also examined whether combinations of vasoactive exposures may interact synergistically to increase risk of ABS and TTLD.

2 ∣. MATERIALS AND METHODS

NBDPS was a multisite, population-based case–control study, designed to investigate more than 30 major structural birth defects and associated genetic and environmental risk factors. From October 1997 to December 2011, birth defects surveillance systems in 10 states across the United States (AR, CA, IA, GA, MA, NJ, NY, NC, TX, and UT) were used to identify cases and nonmalformed controls. Cases include live born, stillborn and, at selected sites, induced abortions. The goal of the study being to identify unknown causes of birth defects, cases of known etiology (e.g., due to chromosomal or genetic abnormalities) were excluded. Control infants were randomly selected among unaffected live births from birth certificates or hospital birth records in the same region, with an estimated delivery date (EDD) during the same period as study cases (Reefhuis et al., 2015; Yoon et al., 2001). Informed consent was obtained from all participants. Overall, 67% of eligible cases and 64% of eligible controls participated in the interviews (Reefhuis et al., 2015). All study protocols and materials were approved by the Institutional Review Boards of participating states and the Centers for Disease Control and Prevention (CDC).

2.1 ∣. Exposures

Mothers of eligible cases and controls were sent introductory packets and invited to complete computer-assisted telephone interviews (CATI), in English or Spanish, between 6 weeks and 24 months after the EDD. Mothers were contacted and interviewed via CATI for approximately 60 min. The interviews included detailed questions on demographics, reproductive factors, behaviors, illnesses, medications for treatment of reported illnesses (e.g., infections, fevers, injuries), and other exposures that occurred 3 months before conception through the end of the pregnancy (Reefhuis et al., 2015). Pregnancy calendars are built into the CATI to assist with accurate reporting of exposures relative to the timing during pregnancy. For medication use, participants were encouraged to report prescription and nonprescription medications, as well as herbal supplements. Information on medications was compiled and coded using the Slone Drug Dictionary, linking drug products to their active ingredients (Kelley, Kelley, Kaufman, & Mitchell, 2003).

Since most structural abnormalities occur in the first trimester of pregnancy, we considered vasoactive exposures occurring during the periconceptional period, defined as 1 month before conception (B1) to the third month of pregnancy (P3). Cigarette smoking was dichotomized into fewer than 15 cigarettes and 15 or more cigarettes per day. Women who reported different average daily amounts of smoking between months during the period of interest were assigned the highest value of reported cigarettes. Reported alcohol consumption was categorized as binge drinking (≥4 drinks/occasion), drinking but not binge drinking (<4 drinks/occasion), and no drinking. Medications of interest (and predominant vasoactivity) included: alpha-adrenergic antagonist action of decongestants as vasoconstrictors, antihypertensive medications and β₂-adrenergic agonists in bronchodilators as vasodilators, and nonsteroidal anti-inflammatory drugs, including aspirin, which can have vasoconstrictive or dilation effects, depending on dose. Participants were considered exposed if they reported any use of the respective medications during the periconceptional period.

Maternal demographic and clinical characteristics were obtained during the telephone interviews. Demographic risk factors included in the descriptive and comparative analyses included were maternal age at delivery (<20, 20–24, 25–29, ≥30 years), years of formal education (<12, 12, 13–15, ≥16 years) and race/ethnicity (non-Hispanic Black, non-Hispanic white, Hispanic, Asian, Native American, and other). Mothers reporting Hispanic ethnicity were classified as Hispanic, regardless of reported race. Anthropometric and reproductive factors of interest included prepregnancy body mass index (BMI) (<18.5, 18.5–<25, 25–<30, ≥30 kg/m² according to the National Institutes of Health categories), number of previous live births (≥1, 0) and whether the pregnancy was intended or not (or unknown/ refused) at the time that it occurred.

2.2 ∣. Outcomes

Case classification has been described in detail elsewhere (Rasmussen et al., 2003; Reefhuis et al., 2015; Yoon et al., 2001). Briefly, cases were classified as isolated, multiple, or complex. The current study only includes isolated cases. Isolated cases referred to cases with single major defects (including single major defects accompanied by minor defects in the absence of a defined syndrome) and multiple major defects in the same organ system. Given the complexity of limb defects classification, isolated cases are presumed to reduce pathogenetic heterogeneity. Isolated ABS cases involved constriction band(s), cord constriction, unusual syndactylies (particularly acrosyndactyly), and amputations only affecting the limbs (i.e., no craniofacial disruption or truncal involvement). ABS cases with reported anomalies consistent with limb–body wall complex were not considered. Limb deficiencies with wrist, mid-arm, and upper arm amputations with nubbins present, without a diagnosis of ABS were classified as “isolated TTLD” if no other major malformations were present. TTLD cases without radiographic evidence of limb reduction were excluded.

2.3 ∣. Statistical analyses

Based on prior findings, we reexamined demographic characteristics, clinical factors, and health behaviors to compare the epidemiologic profiles of ABS and TTLD case groups. Distribution of maternal factors (i.e., age, BMI, education, race/ethnicity, number of previous births, and pregnancy intention) and periconceptional cigarette smoking, alcohol consumption, and vasoactive medication use were computed for cases and controls to characterize higher risk groups. In comparative analyses, maternal age at delivery and education were recategorized into dichotomous variables (age: <25, ≥25 years; education <12 and ≥12 years). For race/ethnicity, mothers reporting Asian or Native American as their racial identity were reclassified as “Other.” For vasoactive medications, mothers whose timing of medication use was reported as unknown or missing were excluded from the comparative analyses. As this study builds on previous analyses using the same data sources, potential covariates included NBDPS center, age at delivery, prepregnancy BMI, education level, race/ethnicity, first birth, and pregnancy intention.

Adjusted odds ratio (aORs) and profile likelihood 95% confidence intervals (CIs) were estimated for case groups with at least five exposed cases using unconditional logistic regression with Firth's penalized likelihood. Firth's penalized-likelihood estimates of beta (β) coefficients and CIs do not assume symmetry of the CI around the β estimate and are more appropriate for small samples (Firth, 1993). Covariates associated with vasoactive exposures and the outcomes of interest, using backward selection with a cut-off level of 0.2, were included in the adjusted model. Participants with missing covariate information were not included in the analyses. To estimate the aORs for the combined exposures, the logistic regression model included recategorized dichotomous terms for smoking (yes vs. no) or alcohol consumption (yes vs. no), regardless of intensity, and reported medications. Estimates were calculated for mothers of cases reporting both smoking and medication use, or alcohol consumption and medication use, using mothers reporting none of the exposures as the reference, among participants with known exposure status for the entire periconceptional period. The aORs were interpreted as relative risks since the outcomes are rare. Analyses were conducted using SAS statistical software version 9.4 (SAS Institute, Inc., Cary, NC).

To explore the possible additive effects of smoking or alcohol and vasoactive medications, we calculated regression coefficients and covariance matrix from the multivariable logistic regression analyses and used an Excel sheet, developed by Andersson, Alfredsson, Kallberg, Zdravkovic, and Ahlbom (2005), to calculate the relative excess risk due to interaction (RERI) and corresponding 95% CIs. If there is no biological interaction, the RERI is expected to equal 0 (Rothman, 2012). Under the assumption of a neutral or monotonic effect of two exposures on all exposed individuals, a RERI > 0 implies synergism (Rothman, Greenland, & Lash, 2008; VanderWeele, 2009; VanderWeele & Robins, 2007). In the absence of a monotonic exposure response, a RERI > 1 can be used to denote possible synergism (VanderWeele, 2009, 2010; VanderWeele & Robins, 2007).

3 ∣. RESULTS

3.1 ∣. Descriptive analyses

After excluding one participant with missing information on EDD, there were 189 ABS cases with limb anomalies classified as “isolated ABS” and 613 TTLD cases classified as “isolated TTLD” among the 32,200 cases and 11,829 controls in the NBDPS from 1997 to 2011. The distribution of the maternal demographic factors (age at delivery, education, race/ethnicity) and clinical factors (prepregnancy BMI, parity, pregnancy intention) are shown in Table 1 for cases and controls. ABS case mothers were younger and less educated than TTLD case mothers and controls. More than half of mothers of ABS cases reported 12 years or less of formal education (51.3%) compared to 41.3% of mothers of TTLD cases and 39.1% of controls. More ABS case mothers identified as Black (23.3%) compared to 8.7% of mothers of TTLD cases and 11.1% of controls. Fewer ABS case mothers identified as Hispanic (19.1%) than TTLD cases (29.2%) and controls (24.6%). Mothers of ABS cases were more likely to be first time mothers but were equally as likely as TTLD case mothers to report an unwanted pregnancy.

TABLE 1.

Descriptive characteristics of mothers of ABS and TTLD cases and nonmalformed controls, National Birth Defects Prevention Study, 1997–2011

	Isolated ABS (n = 189)		Isolated TTLD (n = 613)		Controls (n = 11,829)
Maternal factor	No.	%	No.	%	No.	%
Maternal age, years
<20	34	18.0	63	10.3	1,177	10.0
20–24	65	34.4	140	22.8	2,668	22.6
25–29	41	21.7	170	27.7	3,271	27.7
30+	49	25.9	240	39.2	4,713	39.8
Maternal BMI, kg/m 2
Underweight (<18.5)	10	5.3	34	5.6	599	5.1
Normal weight (18.5–<25)	103	54.5	291	47.5	6,045	51.1
Overweight (25–<30)	30	15.9	143	23.3	2,557	21.6
Obese (30+)	34	18.0	110	17.9	2,074	17.5
Out of range/missing	12	6.4	35	5.7	554	4.7
Maternal education, years
≤12	97	51.3	253	41.3	4,630	39.1
13–15	45	23.8	171	27.9	3,079	26.0
≥16	40	21.2	173	28.2	3,775	31.9
Missing	7	3.7	16	2.6	345	2.9
Maternal race/ethnicity
Black	44	23.3	53	8.7	1,308	11.1
White	93	49.2	350	57.1	6,836	57.8
Hispanic	36	19.1	179	29.2	2,908	24.6
Asian	7	3.7	12	2.0	353	3.0
Native American	1	0.5	2	0.3	51	0.4
Other	7	3.7	17	2.8	366	3.1
Missing	1	0.5	–	–	7	0.1
Number of previous live births
None	110	58.2	266	43.4	4,664	39.4
1+	75	39.7	345	56.3	7,114	60.1
Missing	4	2.1	2	0.3	51	0.4
Pregnancy intention
Wanteda	135	71.4	438	71.5	8,486	71.7
Unwanted	21	11.1	67	10.9	1,115	9.4
Refused/unknown	0	–	0	–	14	0.1
Missing	33	17.5	108	17.6	2,214	18.7

Abbreviations: ABS, amniotic band syndrome; TTLD, terminal transverse limb deficiencies.

^aIncludes “mistimed” and “did not care” responses.

The distributions of maternal vasoactive exposures are shown in Table 2. More ABS case mothers (26.4%) were cigarette smokers than TTLD case mothers (20.1%) or 17.5% of controls. About 43.4% of ABS cases reported alcohol consumption compared to 34.4% of TTLD cases and 35.7% of controls. ABS case mothers were also more likely to report binge drinking (18.0%) than both TTLD case mothers (13.1%) and controls (12.1%). A greater proportion of TTLD case mothers reported use of decongestants compared to mothers of ABS cases (15.3 vs. 9.0%). ABS case mothers were more likely to use bronchodilators (5.8 vs. 2.8%) and aspirin (9.0 vs. 5.1%) than TTLD case mothers. Use of nonaspirin NSAIDS was almost identical across case groups (26.9 vs. 26.5%) and higher compared to controls (22.5%). Frequency of bronchodilator (3.1%) and aspirin use in controls (4.2%) was most similar to TTLD cases while frequency of decongestant (10.1%) was most similar to ABS cases. None of the ABS case mothers and less than 1% of TTLD cases and controls reported use of antihypertensive medications.

TABLE 2.

Periconceptional health behaviors and exposures among mothers of ABS and TTLD cases and nonmalformed controls, National Birth Defects Prevention Study, 1997–2011

	Isolated ABS (n = 189)		Isolated TTLD (n = 613)		Controls (n = 11,829)
Periconceptional exposures	No.	%	No.	%	No.	%
Maternal smoking
≥15 cigarettes/day	15	7.9	35	5.7	580	4.9
<15 cigarettes/day	35	18.5	88	14.4	1,495	12.6
No smoking	132	69.8	477	77.8	9,454	79.9
Missing/unknown	7	3.7	13	2.1	300	2.5
Alcohol drinking
Binge drinking (≥4 drinks)	34	18.0	80	13.1	1,431	12.1
Drinking but not binge drinking	48	25.4	131	21.4	2,788	23.6
No drinking	98	51.9	384	62.6	7,210	61.0
Missing/unknown	9	4.8	18	2.9	400	3.4
Medications
Decongestants	17	9.0	94	15.3	1,193	10.1
Bronchodilators	11	5.8	17	2.8	364	3.1
Antihypertensives	0	–	2	0.3	81	0.7
Nonaspirin NSAIDs	50	26.5	165	26.9	2,658	22.5
Aspirin	17	9.0	31	5.1	492	4.2

Abbreviations: ABS, amniotic band syndrome; NSAIDs, nonsteroidal anti-inflammatory drugs; TTLD, terminal transverse limb deficiencies.

3.2 ∣. Comparative analyses

aORs for maternal vasoactive exposures and health behaviors are presented in Table 3. There was some evidence of increased risk of both ABS and TTLD for cigarette smoking, with a suggested dose–response pattern for greater smoking levels in both groups. Increased risk of ABS was noted for both binge drinking (aOR 1.3, 95% CI 0.7–2.1), and drinking without binging (aOR 1.6, 95% CI 1.0–2.3). This relationship was absent in the TTLD case group. For ABS, aORs were close to the null for decongestants (1.0, 95% CI 0.6–1.8) and nonaspirin NSAIDs (1.3, 95% CI 0.8–1.9) and elevated for use of bronchodilators (1.8, 95% CI 0.8–3.3) and aspirin (2.2, 95% CI 1.2–3.9). In contrast, aORs for TTLD were elevated for decongestants (1.7, 95% CI 1.3–2.1) and nonaspirin NSAIDS (1.3, 95% CI 1.1–1.7) and close to the null for bronchodilators (0.9, 95% CI 0.5–1.5) and aspirin (1.2, 95% CI 0.8–1.8). Due to small sample size, associations with antihypertensives could not be estimated.

TABLE 3.

aORs estimates (and 95% CI) for periconceptional vasoactive exposures among mothers of ABS and TTLD cases compared to controls, National Birth Defects Prevention Study, 1997–2011

	Isolated ABS (n = 141)		Isolated TTLD (n = 458)
Periconceptional exposures	aORa	95% CI	aORa	95% CI
Maternal smoking
≥15 cigarettes/day	1.5	(0.7, 2.8)	1.3	(0.9, 2.0)
<15 cigarettes/day	1.3	(0.8, 2.0)	1.2	(0.9, 1.6)
No smoking	Ref		Ref
Alcohol consumption
Binge drinking (≥4 drinks)	1.3	(0.7, 2.1)	0.9	(0.7, 1.2)
Drinking but not binge drinking	1.6	(1.0, 2.3)	0.9	(0.7, 1.1)
No drinking	Ref		Ref
Medications b
Decongestants	1.0	(0.6, 1.8)	1.7	(1.3, 2.1)
Bronchodilators	1.8	(0.8, 3.3)	0.9	(0.5, 1.5)
Antihypertensivesc	–	–	–	–
Nonaspirin NSAIDs	1.3	(0.8, 1.8)	1.3	(1.1, 1.7)
Aspirin	2.2	(1.2, 3.9)	1.2	(0.8, 1.8)

Note: Adjusted for age, race/ethnicity, education, previous live births, and study center. Abbreviations: ABS, amniotic band syndrome; aOR, adjusted odds ratio; CI, confidence interval; NSAIDs, nonsteroidal anti-inflammatory drugs; ORs, odds ratio; TTLD, terminal transverse limb deficiencies.

^aModels exclude women who were missing education, race/ethnicity, and number of previous live births (refer to Table 1).

^bWomen whose timing of use was reported as unknown or missing were also excluded.

^cAdjusted ORs for cell sizes less than 5 were not calculated, represented by –.

The results of the multivariable analysis for the combination of vasoactive exposures among cases with known exposure status for smoking, drinking, and medications during the entire periconceptional period are presented in Table 4. Based on 7 exposed cases and 123 exposed controls, the combination of smoking and aspirin was associated with increased risk of ABS (aOR 3.3, 95% CI 1.3–7.1). Combined exposure to smoking and bronchodilators, also suggested evidence of elevated risk of ABS (aOR 2.7, 95% CI 0.9–6.7) among five cases. Marked increases in risk of ABS were also associated with any combination of reported alcohol consumption and medications, most notably, a greater than threefold increased risk (aOR 3.4, 95% CI 1.4–7.5) in eight cases reporting alcohol consumption and use of bronchodilators, as well as an almost fourfold increased risk in 10 cases reporting both alcohol consumption and use of aspirin (aOR 4.5, 95% CI 1.8–10.5) during the periconceptional period. Increased risk of TTLD was only observed in cases reporting a combination of cigarette smoking and decongestants (aOR 2.3, 95% CI 1.4–3.6). The largest RERI values were observed for ABS cases exposed to bronchodilators and smoking (1.4) or alcohol (2.0), as well as aspirin and smoking (1.4). However, the lower bound of the 95% CIs for the RERI values of all the combinations of exposures included zero and the absence of synergism could not be ruled out (Table 5).

TABLE 4.

Exposure to cigarette smoking or alcohol, and medications among mothers of ABS and TTLD cases, National Birth Defects Prevention Study, 1997–2011

	Isolated ABS (n = 78)		Isolated TTLD (n = 254)		Controls
Combined exposuresa,b	n exp.	aOR (95% CI)	n exp.	aOR (95% CI)	n exp.
Cigarette smoking + medications
Smoking + decongestants	20	1.1 (0.3, 3.1)	41	2.3 (1.4, 3.6)	693
Smoking + bronchodilatorsc	5	2.7 (0.9, 6.7)	3	–	92
Smoking + nonaspirin NSAIDs	12	1.1 (0.5, 2.5)	43	1.5 (1.0, 2.2)	609
Smoking + aspirin	7	3.3 (1.3, 7.1)	9	1.4 (0.7, 2.7)	123
Alcohol consumption + medications
Alcohol + decongestants	13	1.5 (0.7, 2.9)	32	1.0 (0.7, 1.6)	539
Alcohol + bronchodilators	8	3.4 (1.4, 7.5)	6	0.6 (0.2, 1.3)	151
Alcohol + nonaspirin NSAIDs	33	1.5 (0.8, 3.1)	74	0.9 (0.6, 1.3)	1,316
Alcohol + aspirin	10	3.7 (1.6, 7.8)	16	1.0 (0.6, 1.7)	231

Note: Adjusted for age, race/ethnicity, education, previous live births and study center. Abbreviations: ABS, amniotic band syndrome; aOR, adjusted odds ratio; CI, confidence interval; NSAIDs, nonsteroidal anti-inflammatory drugs; ORs, odds ratios; TTLD, terminal transverse limb deficiencies.

^aWomen whose timing of use was reported as unknown or missing were also excluded.

^bModels exclude women who were missing education, race/ethnicity, and number of previous live births (refer to Table 1).

^cORs for cell sizes less than 5 were not calculated, represented by –.

TABLE 5.

Relative risk due to interaction for combinations of vasoactive exposures among mothers of ABS and TTLD cases, National Birth Defects Prevention Study, 1997–2011

	Isolated ABS (n = 78)		Isolated TTLD (n = 254)
Combined exposuresa,b	aOR	RERI (95% CI)	aOR	RERI (95% CI)
Cigarette smoking + medications
Smoking + decongestants	1.1	−0.7 (−2.6, 1.3)	2.3	0.7 (−0.4, 1.8)
Smoking + bronchodilatorsc	2.7	1.4 (−1.4, 4.2)	–	–
Smoking + nonaspirin NSAIDs	1.1	−1.7 (−3.8, 0.4)	1.5	0.0 (−0.9, 0.9)
Smoking + aspirin	3.3	1.4 (−1.5, 4.3)	1.4	0.3 (−1.1, 1.7)
Alcohol consumption + medications
Alcohol + decongestants	1.5	0.6 (−0.5, 1.7)	1.0	−0.6 (−1.5, 0.2)
Alcohol + bronchodilatorsc	3.4	2.0 (−0.8, 4.9)	0.6	0.0 (−0.8, 0.7)
Alcohol + nonaspirin NSAIDs	1.5	0.1 (−1.0, 1.3)	0.9	0.0 (−0.5, 0.4)
Alcohol + aspirin	3.7	0.5 (−2.6, 3.6)	1.0	0.2 (−0.7, 1.0)

Note: Adjusted for age, race/ethnicity, education, previous live births, and study center. Abbreviations: ABS, amniotic band syndrome; aOR, adjusted odds ratio; BMI, body mass index; CI, confidence interval; NSAIDs, nonsteroidal anti-inflammatory drugs; ORs, odds ratios; RERI, relative excess risk due to interaction; TTLD, terminal transverse limb deficiencies.

^aModels exclude women who were missing BMI, education, race/ethnicity, number of previous live births, and pregnancy intent (refer to Table 1).

^bWomen whose timing of use was reported as unknown or missing were also excluded.

^cORs for cell sizes less than 5 were not calculated.

4 ∣. DISCUSSION

Based on clinical (Hoyme, Higginbottom, & Jones, 1981; McGuirk et al., 2001) and animal studies (Brent & Franklin, 1960; Webster & Brown-Woodman, 1990; Webster, Lipson, & Brown-Woodman, 1987), vascular disruption has been proposed as a causal mechanism for limb reduction defects. Vasoactive exposures have been examined in previous studies as possible markers for vascular disruption theory. These exposures include cigarette smoking, alcohol consumption, and vasoactive medications (Caspers Conway, Romitti, Holmes, Olney, & Richardson, 2014; Werler et al., 2004; Werler et al., 2009; Werler, Louik, & Mitchell, 2003; Werler, Sheehan, & Mitchell, 2003). We sought to build on previous analyses of NBDPS data to examine whether there is evidence for a vascular disruption pathogenesis of ABS and TTLD, and whether the risk of these birth defects differed by vasoactive property.

These findings support previous conclusions that TTLD and ABS may be different entities (Werler et al., 2009). Our results were generally different for ABS compared to TTLD, with strongest evidence of increased risk of ABS associated with vasodilation such as exposure to alcohol and bronchodilators, while risk of TTLD was associated with vasoconstrictive exposures such as cigarette smoking and decongestant use (Werler et al., 2009). Specifically, the combination of smoking and decongestants, both vasoconstrictive exposures, increased the risk of TTLD by more than twofold, while exposures to bronchodilators and aspirin in combination with either smoking or alcohol tripled, in the case of alcohol and aspirin—nearly quadrupled, the risk of ABS.

The positive association between maternal smoking in pregnancy and ABS is consistent with the previous NBDPS analyses (Werler et al., 2009) and two other studies (Czeizel, Vitez, Kodaj, & Lenz, 1993; Wasserman, Shaw, O'Malley, Tolarova, & Lammer, 1996), but is in contrast to null findings from another study (Werler, Louik, & Mitchell, 2003; Werler, Sheehan, & Mitchell, 2003). Our findings of slightly elevated risk of TTLD associated with cigarette smoking are inconsistent with the previous analyses of these data for years 1997–2004 (Werler et al., 2009), but are consistent with literature on smoking and limb reduction malformations (Caspers et al., 2013; Czeizel, Kodaj, & Lenz, 1994; Kallen, 1997; Wasserman et al., 1996). Additionally, maternal cigarette smoking and increased risk of other deficiencies attributed to vascular disruption (Baldacci et al., 2020; Interrante et al., 2017; Lam & Torfs, 2006; Skarsgard et al., 2015), as well increased risks due to the combination of cigarette smoking and vasoactive medications (Werler et al., 2004; Werler, Louik, & Mitchell, 2003; Werler, Sheehan, & Mitchell, 2003) have also been reported.

Other studies have examined maternal alcohol use in relation to the broader group of limb reduction deficiencies (Aro, Haapakoski, & Heinonen, 1984; Caspers Conway et al., 2014; Froster & Baird, 1992; Martinez-Frias, Bermejo, Rodriguez-Pinilla, & Frias, 2004; Shaw et al., 2002) and reported both increased and null associations for various measures of alcohol intake. The lack of consistent findings could be attributed to differences in exposure and outcome definitions (Aro et al., 1984; Froster & Baird, 1992; Martinez-Frias et al., 2004; Shaw et al., 2002). Additionally, the threshold at which alcohol becomes a teratogen remains unknown (Henderson, Gray, & Brocklehurst, 2007). Our estimates of the associations between exposure to vasoactive medications and both ABS and TTLD were fairly consistent with the previous studies (Werler et al., 2009). Our observed elevated risks of ABS associated with use of bronchodilators are stronger than reported in the earlier NBDPS analysis (Werler et al., 2009). Bronchodilators can have both vasoconstrictive and vasodilating effects, but the latter is predominant. Interestingly, odds ratios were strongest for bronchodilator use in combination with either vasoconstrictive cigarette smoking or vasodilating alcohol intake. Leaving vasoactivity aside, it is also possible that other factors related to social disadvantage confound these elevated odds ratios, as alcohol intake in combination with each of the other vasoactive medications (decongestants, NSAIDs, and aspirin) were also positively associated with ABS.

Demographic, reproductive and clinical factors, as well as health behaviors associated with increased risk of TTLD had no apparent association with ABS, consistent with previous literature (Garza et al., 1988; Werler et al., 2009). ABS case mothers were more likely to be younger, identify as Black, have lower levels of education or report first live birth compared to TTLD mothers. Young age, lower education, and first births have all been linked to amniotic bands (Werler et al., 2009; Werler, Louik, & Mitchell, 2003; Werler, Sheehan, & Mitchell, 2003), consistent with our observations. Again, this suggests social marginalization or factors related to reproductive immaturity as etiologic factors for ABS.

Strengths of our study include the use of the NBDPS, the largest U.S. population-based study of birth defects, with specific case definitions and classifications by clinical geneticists. Multivariable logistic regression models adjusted for several potential confounders including maternal age, study location and number of previous live births, reducing the likelihood that observed results were due to confounding. Our results are consistent with previous analyses on vasoactive exposures and risks of the ABS and TTLD. Additionally, we evaluated the potential additive effects of combinations of vasoactive exposures, computed RERIs and 95% CI.

Several limitations should be considered when interpreting our results. First, our findings on vasoactive medications are limited by small sample size. We could not evaluate associations with antihypertensives. Second, we lacked information on indication and dose of medications. Specifically, NSAIDs and aspirin have dilating properties that are dose dependent and we were unable to assess whether the observed increased risk of ABS and aspirin differed by ingested dose. Third, information on exposures is collected retrospectively and exposures may have occurred as much as 2 years prior to the interview. Recall is subject to misclassification. Particularly, episodic exposures such as medications may be reported inaccurately. However, our results are consistent with prior research and previous literature that differences in accuracy of maternal recall in cases and controls are unlikely to create spurious associations or bias results away from the null (Drews, Kraus, & Greenland, 1990). Our studies support previous conclusions that ABS and TTLD are two different entities and suggest that TTLD may be associated with vasoconstriction, while ABS may be associated with vasodilation. There is possible synergism due to multiple vasoconstrictive or vasodilating exposures; however, numbers from this study were insufficient to properly examine combined exposures. Further investigation in the effect of multiple exposures with the same vasoactive property will help clarify whether specific interactions of exposures are increasing the risks of these and other specific congenital malformations. Additionally, factors associated with social disadvantage and reproductive immaturity may also play a role in the etiology of ABS. Future studies could consider the role of amnion rupture on other pregnancy outcomes such early preterm delivery.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.