Geoepidemiology of systemic sclerosis

Abstract

Systemic sclerosis (SSc) is a rare and potentially severe connective-tissue disease, characterized by skin fibrosis and involvement of internal organs. Because of its rarity and heterogeneous clinical presentation, reliable epidemiological studies on SSc have been particularly difficult to carry out. SSc prevalence is estimated between 3 and 24 per 100,000 population and appears to be higher in North America and Australia as compared to Europe and Japan. Incidence estimates have significantly increased between the fifties and the eighties, but this could result from greater physician awareness of the disease and more reliable ascertainment methods. Risk factors for SSc include female sex and African origin. Reports of sporadic clusters of higher prevalence also suggest environmental risk factors. In particular, silica and solvents exposure has been associated with SSc by several rigorous case–control studies. The ten-year cumulative survival of SSc has improved significantly from 50% in the seventies to over 70% at the present time. Pulmonary fibrosis and pulmonary arterial hypertension are now the two main causes of death. Diffuse cutaneous forms, as well as cardiac, pulmonary, and renal involvement are independent risk factors for SSc-related mortality.

Introduction

Systemic sclerosis (SSc) is a connective-tissue disease characterized by excessive collagen deposition in the dermis and internal organs and by vascular hyper-reactivity and obliterative microvascular phenomena. Patients with SSc are usually classified into two main groups, according to the extent of skin involvement: limited SSc (lSSc), with skin involvement essentially limited to the hands and face; and diffuse SSc (dSSc), with skin involvement proximal to the elbows and knees. In patients with lSSc, visceral involvement is rare, with the exception of patients in whom pulmonary arterial hypertension, interstitial lung disease and/or bowel involvement eventually develops. Patients with dSSc experience visceral involvement, including renal crisis, interstitial lung disease, pulmonary arterial hypertension, heart and gastrointestinal tract involvement.

The study of SSc represents a challenge for epidemiologists, because of the disease's rarity and the large spectrum of clinical manifestations and severity, with various diagnosis and classification criteria over time. Researchers have thus encountered many difficulties to carry out reliable surveys, since the disease is too rare to be detected by cross sectional population surveys. Therefore, epidemiologic studies of SSc are characterized by a particularly large variability in estimates of prevalence, incidence, risk factors and mortality rates, partly due to variability in the design and methods used to calculate these parameters. However, identifying changes in the epidemiological pattern of the disease may help in finding key determinants of risk and causal factors.

The absence of common definition criteria for SSc made epidemiologic studies very difficult to compare until the eighties. Therefore, prevalence and incidence estimates from the earlier surveys must be interpreted cautiously. In 1980, the American Rheumatism Association (ARA) classification, that was developed to distinguish SSc from other connective-tissue diseases, has provided a common definition of the disease [1], allowing more homogeneous epidemiological studies. Because it was designed for research applications and not for clinical diagnosis, it has been criticized for its low sensitivity in identifying early disease and milder forms of systemic scleroderma such as CREST syndrome. The name CREST is an acronym for the cardinal clinical features of the syndrome (calcinosis, Raynaud's phenomenon (RP), esophageal dysmotility, sclerodactyly, and telangiectasia). Thus, many authors also included patients with CREST features in their surveys. In 1988, LeRoy and colleagues proposed a new classification, splitting the disease's spectrum into two main forms: lSSc and dSSc. Finally, in 2001, Leroy and Medsger proposed revised criteria, including recent technological improvement, such as autoantibodies (anticentromere, topoisomerase, fibrillarin, PM-Scl, RNA polymerases I and III) and nailfold capillaroscopy for the diagnosis of early SSc [2] that distinguishes lSSc, with no detectable skin involvement; limited cutaneous SSc (lcSSc), with skin involvement essentially limited to the hands and face and diffuse cutaneous SSc (dcSSc), with skin involvement proximal to the elbows and knees. This classification is now used by most epidemiological studies of SSc, although some patients with dermatomyositis or mixed connective-tissue diseases might fulfil these criteria. Some authors also include “scleroderma spectrum disorders” encompassing dSSc and lSSc, CREST syndrome, SSc sine scleroderma, undifferentiated connective-tissue diseases (UCTD) and overlap syndromes. The resulting surveys hence provide higher prevalence/incidence estimates.

The method of case ascertainment differs widely among studies. Early studies usually used a single source of information. One of the most frequent methods was to review discharge diagnosis of SSc in major hospitals, using international disease classification codes, assuming that all patients were hospitalized because SSc is a severe disease. Such studies neglected outpatients that were not hospitalized and therefore underestimated the prevalence of the disease, especially for mild forms. Chandran et al. estimated that the inpatients-to-outpatients ratio was 1/1 in 3 hospitals of South Australia [3]. If this estimation is correct, the true prevalence of SSc is twofold the prevalence estimated by hospital based surveys. Moreover, single case ascertainment, including population-based administrative data (physician billing) and hospitalization databases do not have a sufficient sensitivity to provide reliable estimates of incidence/prevalence for SSc on their own [4]. Nowadays, most studies use multiple sources of ascertainment. This methodological improvement has been facilitated by the creation of computerized databases (ex: SASR in South Australia in 1993), patients support groups and access to public health insurance databases. In addition, statistical methods are now commonly employed, that take into account the incomplete case finding arising from a single source of ascertainment and correct the calculated prevalence rate. The most widely used is the so-called method of capture–recapture [5], [6], [7], that takes advantage of duplicate information derived from multiple sources of case ascertainment, to evaluate the degree of overlap among sources, estimate the number of cases missed by one source and hence estimate the true number of cases in a given area. On the opposite, all methods may overestimate the prevalence of the disease because of incorrect diagnoses of SSc in databases (for example, patients with mixed connective-tissue disease (MCTD) are often incorrectly encoded as SSc). Ideally, diagnoses should be confirmed by manual consultation of medical charts, but this is usually not possible for all cases. A good alternative is to estimate the proportion of reliable diagnoses using the available medical charts, and correct the number of the remaining unchecked cases accordingly [7].