PATIENTS AND METHODS
- Top of page
- PATIENTS AND METHODS
Cases included all women diagnosed with SSc and clinically followed up at the Institute of Internal Medicine of the University Hospital of Verona between January 1, 1997 and June 30, 1999. The diagnosis of SSc was made according to the criteria of LeRoy et al (6). Controls were selected among patients hospitalized in the same period at the Institute of Orthopedics in the same University Hospital. Cause of admission for controls included fracture, dislocation, and sprain (International Classification of Diseases, Ninth Revision [ICD-9] codes 800–829, 830–839, and 840–848, respectively) (n = 59, or 39%), disorders of the joint (ICD-9 codes 717.0–717.9, 718.0–718.9, 719.0–719.9, 726.0–726.9, 727.0–727.9) (n = 32, or 21%), carpal tunnel syndrome (ICD-9 code 354.0) (n = 23, or 15%), osteoarthrosis (ICD-9 codes 715.0–715.9) (n = 16, or 10%), acquired deformities (ICD-9 codes 736.0–736.9, 737.0–737.9) (n = 9, or 6%), osteoporosis (ICD-9 code 733.0) (n = 3, or 2%), bone infections (ICD-9 codes 730.0–730.9) (n = 3, or 2%), disorders of the muscle, ligament, and fascia (ICD-9 codes 728.0–728.9) (n = 2, or 1%), and benign (ICD-9 codes 213.0–213.9, 214.0, 215.0–215.9) (n = 3, or 2%) and malignant (ICD-9 codes 170.0–170.9) (n = 3, or 2%) neoplasms. Frequency matching by sex and age group was applied to the selection of controls.
The participation rate was 100% among cases and 98% among controls, leading to a total of 46 cases, 42 with diffuse SSc and 4 with the limited cutaneous variant of the disease (ICD-9 code 710.1), and 153 controls. All of the study subjects were white.
Each study participant completed an in-person interview, which was conducted by a trained interviewer. A structured questionnaire was used to investigate in detail the reproductive history of the subject, with questions on the number of pregnancies and abortions (either spontaneous or induced), age at first pregnancy, and use of oral contraceptives. The questionnaire also provided information on date of birth, education, presence of silicone breast implants or cosmetic surgery, presence of pets in the household, lifetime smoking, alcohol consumption, occupational history, including job titles, industries, and duration of employment for each job title, as well as employment in occupations carrying the potential for exposure to organic solvents (aromatic and aliphatic hydrocarbons), silica, hand-arm vibration, and certain chemicals (vinyl chloride, formaldehyde, epoxy resins). Participants were also asked to indicate their past use of selected drugs (bleomycin, pentazocine, appetite suppressants, L-tryptophan, carbidopa, and cocaine) for which suggestions of a role in SSc causation have been raised (7).
To estimate the relative risk for SSc according to potential predictors, the odds ratio (OR) and its 95% confidence interval (95% CI) were calculated using unconditional logistic regression. In analyses according to reproductive history, either nulliparous women or women who had never been pregnant were used as the reference category. Age at first pregnancy was classified in 2 categories using 25 years (the mean value among controls) as the cut-off. Women who did not report any abortion were the reference category to measure the relative risk according to history of abortion. Occupation, occupational exposures, education, presence of pets in the household, past use of selected drugs, lifetime smoking, and alcohol consumption were evaluated in multivariate models. However, these terms were excluded from the final model because their inclusion did not change the relationship between SSc and the variables linked to reproductive history.
- Top of page
- PATIENTS AND METHODS
The distribution of the 46 cases and 153 controls according to personal and demographic characteristics, variant of disease, and age at diagnosis is shown in Table 1. Cases were older than controls and had a comparable level of education. The distribution according to area of residence was similar among cases and controls, since 80% of cases and 77% of controls were residents in the province of Verona. Among both cases and controls, the majority of subjects were nonsmokers, while regular alcohol drinking was more frequent in the controls (38%) than in the cases (7%). Presence of silicone breast implants and cosmetic surgery were reported by only 4 controls and none of the cases. Only 4 of the case subjects were diagnosed with the limited variant of SSc.
Table 1. Distribution of cases of scleroderma and controls according to selected covariates
| ||Cases (n = 46)||Controls (n = 153)|
| Primary school||24||52||70||46|
| Middle school||11||24||47||31|
| High school||8||17||31||20|
|Past use of scleroderma-related drugs|
|Presence of silicone breast implants|
|Variant of disease|
The diagnosis of SSc was made more often after childbearing age: 17% of cases were diagnosed between ages 45–49 years and 56% at age 50 or more. The mean age at diagnosis was 53 years (range 27–73 years). In all cases, the first pregnancy preceded the diagnosis of SSc; for 2 women, the time interval between the first pregnancy and the diagnosis was shorter than 10 years (1 year and 9 years, respectively), for 5 women, this interval spanned 10–19 years, for 13 women, 20–29 years, and for 12 women, ≥30 years.
The relationship between SSc and reproductive history is displayed in Table 2. Compared with nulliparous women, parous women were at reduced risk of developing SSc (age-adjusted OR 0.3, 95% CI 0.1–0.8) and the risk showed a decrease with an increasing number of children. The age-adjusted OR was 0.6 (95% CI 0.2–1.7) for those having had 1 child, 0.3 (95% CI 0.1–0.7) for those having had 2 children, and 0.3 (95% CI 0.1–0.8) for those having had 3 or more children. Abortive pregnancies were also inversely related to SSc; women who had experienced an abortion had an OR of 0.5 (95% CI 0.2–1.5) compared with women reporting no history of abortion. When we considered all the pregnancies experienced by a woman, either complete or abortive, the age-adjusted OR was 0.3 (95% CI 0.1–0.7).
Table 2. Odds ratio (OR) and 95% confidence interval (95% CI) for the risk of scleroderma according to reproductive covariates
| ||Cases (n = 46)||Controls (n = 153)||Univariate||Age-adjusted|
| ||No.||%||No.||%||OR||95% CI||OR||95% CI|
|Number of children|
|Any pregnancy (complete + abortive)|
|Age at first pregnancy|
| 25+ years||18||39||58||38||0.5||0.2–1.2||0.4||0.2–1.0|
| <25 years||13||28||71||46||0.3||0.1–0.8||0.3||0.1–0.7|
|Current oral contraceptives use|
Compared with women who had never been pregnant, women who had the first pregnancy at age 25 years or older had an age-adjusted OR of 0.4 (95% CI 0.2–1.0), whereas the age-adjusted OR was 0.3 (95% CI 0.1–0.7) for women younger than 25 years at their first pregnancy. When that analysis was restricted to women who had experienced at least 1 pregnancy, an older age at first pregnancy was associated with a slight increase in risk (age-adjusted OR 1.5, 95% CI 0.7–3.5).
Use of oral contraceptives was not associated with an increase in risk for SSc. Compared with women who were taking oral contraceptives, the age-adjusted OR for nonusers was 1.2 (95% CI 0.5–3.2). This estimate was not modified when we considered only premenopausal women younger than 50 years (age-adjusted OR 1.2, 95% CI 0.3–4.6; 12 cases with 4 users and 8 nonusers, and 66 controls with 25 users and 41 nonusers).
When we restricted the analyses to the 42 women with diffuse SSc, all of these results were confirmed (data not shown).
- Top of page
- PATIENTS AND METHODS
This study showed a reduced risk of SSc among parous women when compared with nulliparous women, and an inverse association between the risk for SSc and the number of children. When we considered any pregnancy, either complete or abortive, the reduction in risk among women who had been pregnant, compared with women who had not, was of the same magnitude. No relationship was seen between current use of oral contraceptives and the occurrence of SSc. The reduction in risk showed by women who had an abortion enhances the internal coherence of our findings.
Scleroderma is a rare autoimmune disease of unknown etiology, with a marked excess in incidence among women. This excess is particularly relevant after childbearing age, when the incidence shows a peak (1). Due to the rarity of the disease, its insidious onset, and the possible overlap of its clinical and laboratory features with those of other rheumatic or autoimmune diseases, few epidemiologic studies have addressed and quantified the role of potential risk factors for SSc. Most of these studies have, indeed, focused on occupational exposures (8). The hormonal status and its modifications or factors and events related to the pregnancy have been suggested to explain the predominance of the disease in women and its pattern according to age (1).
In studies of molecular biology, it appears that cell exchange occurs between the mother and the fetus during normal human pregnancy and that fetal stem cells can survive in maternal blood and tissues for many years after the pregnancy (9). This presence of small populations of cells from another individual, called microchimerism, has been hypothesized to be involved in the pathogenesis of some autoimmune diseases, notably SSc. Immunobiologic modifications during the pregnancy would thus be involved in SSc causation. Fetal–maternal HLA compatibility would also play a role in this model; fetal HLA-compatible cells might not be rejected from the mother's immune system and would therefore persist after the delivery (2). In a study on 37 women with SSc and 42 controls, Artlett et al (10) reported a higher frequency of HLA–class II compatibility between the study subject and her children or her mother among the cases than among the controls. The authors, however, considered compatibility with both the subject's mother and the subject's children, and since no data were shown for controls, the reader is not able to determine whether a significantly higher frequency of subject–children compatibility was found among the cases. Moreover, 31 of 42 controls were women with pemphigoid gestationis, a condition described by the authors as displaying “provocative similarities” to SSc, including an HLA association (DR3/DR4), and female predominance. These observations cast doubts on the validity of the choice of the controls in that study.
Nelson et al (3) compared 17 SSc patients with 23 healthy controls who had had at least 1 son and found that the concentration of male DNA in the whole blood was higher in cases than in controls. In the same study, previous pregnancies with HLA-compatible children occurred more frequently among 21 women with SSc than among 32 controls. The 2 case–control groups overlapped only in part and correlation between microchimerism and HLA compatibility could not be assessed. In another study from the same group (5), the presence of Y chromosome–specific DNA sequences in peripheral blood mononuclear cells was more frequent in case than in control women. In a subpopulation of 17 women (6 SSc cases) with complete family HLA information, the persistence of microchimerism in peripheral blood mononuclear cells was not related to HLA compatibility between the women and their previously born sons. Artlett et al (4) found a higher frequency of Y chromosome sequences in DNA from peripheral blood cells and skin lesions in women with SSc than in control women. However, the assessment of the reproductive history was incomplete and a misclassification of the “exposure” might have biased the results.
All these studies are based on a small number of subjects and chance cannot be ruled out in the explanation of the results. Moreover, if the microchimerism would occur more frequently among women with SSc, then the biologic meaning of this finding would not be easily explained. In fact, the persistence of the fetal non-self cells could also be a consequence of the primitive impairment of the immune system that would eventually lead to the clinically evident disease.
No epidemiologic studies have evaluated the effect of previous pregnancies on the risk of developing SSc, and the immunobiologic hypothesis has never been assessed in an epidemiologic study. Our results indicating a reduced risk of SSc in women who had been pregnant, compared with women who had not, seem to indicate that pregnancy is not a risk factor for SSc. We do not know the percentage of fetal–maternal HLA compatibility in our cases and controls, and therefore we are not able to evaluate if HLA compatibility modifies the effect of pregnancy on the risk. The inverse relationship between the risk of SSc and pregnancy can be alternatively explained if the disease itself would impair fertility or induce the affected women to avoid pregnancies. However, since among all cases, the first pregnancy preceded the diagnosis, it seems unlikely that such an explanation would account for the inverse relationship observed.
Subclinical alterations occurring before clinical disease might affect fertility or the course and outcome of the pregnancy because of the insidious onset and the microvascular alterations that are characteristic of SSc. Women with SSc have indeed been reported to have a higher risk of infertility or miscarriages prior to the diagnosis. Focusing on the reproductive history of the women before the first symptoms of SSc, a case–control study (11) found a nonsignificant excess relative risk of infertility, defined as a self-reported difficulty in conception and the failure to achieve a successful pregnancy by the age of 35, among women with SSc compared with friend controls. A significant increase in the risk of spontaneous abortion was also found. In another case–control study (12), the same investigators found a higher risk of delay in conception and infertility among women with SSc when compared with healthy women, but failed to find an excess risk when compared with a control group of women with primary Raynaud's phenomenon. We do not have information on infertility in our study group. It therefore cannot be ruled out that a lowered fertility due to subclinical disease explains the reduced risk of SSc among the women who had been pregnant.
Owing to the lack of individual matching in the study design, truncation of exposure (the reproductive history) of controls could not be performed on the basis of the diagnosis of the cases. We recognize this limitation, but we suggest that the dichotomous classification of parous/nulliparous would not have changed if truncation of exposure period was applied.
We considered the relevant exposure to begin with the first pregnancy, which for all cases, preceded the diagnosis. Since we do not know the exact timing of the abortions or subsequent pregnancies with respect to the disease onset, we could not limit the analyses to events known to have occurred prior to disease onset. However, since the distribution of cases and controls in the 2 mutually exclusive categories, parous/nulliparous, would not have changed on the basis of the timing between subsequent pregnancies or abortions and the diagnosis, we believe that our results are not affected by any misclassification of exposure. Moreover, considering the age at diagnosis, we believe that, in addition to the first pregnancy, the majority of subsequent pregnancies are also very likely to have occurred prior to the diagnosis.
The reproductive history can be related to social status. Since the Institute of Internal Medicine of the University Hospital of Verona is a referral center for rheumatic diseases for the surrounding area of Northeastern Italy, it is unlikely that an elevated number of incident diagnoses of SSc in the population served by this clinic had been missed. Since, in Italy, access to medical care is provided by the National Health Service, we believe that patients had referred themselves to the University Hospital independent of their social status. Moreover, the education level of the cases and controls was comparable. The high comparability of controls and cases according to the residence area indicates that controls stemmed from the same general population in which cases occurred. Therefore, it seems unlikely that a bias in the selection of cases and controls would have affected our results.
According to our data, education and other explanatory variables do not appear to confound the relationship of SSc with reproductive status, nor do other variables have an effect, such as presence of pets in the household, past use of selected drugs, lifetime smoking, and alcohol consumption.
Since the onset of SSc is insidious and preclinical disease might reduce the fertility or increase the risk of spontaneous abortion, it is not easy to ascertain the relationship with reproductive history and the effect of parity on the risk. Showing an inverse association of the risk of SSc with a history of pregnancy, our results seem not to support the hypothesis that immunobiologic modifications during any pregnancy explain the peak of incidence of the disease observed among women, especially after their childbearing age. Further study, however, is needed to evaluate the role of HLA-compatible pregnancies on the risk of subsequent SSc development. Moreover, other factors, for instance, age-related hormonal changes, deserve attention because they might be involved in SSc causation or increasing the susceptibility to SSc in women.