1. Top of page
  2. Abstract


To determine the incidence of cancer in patients with scleroderma (systemic sclerosis) and to compare those rates with cancer rates in the local population.


Cancer risk in scleroderma patients in the Detroit metropolitan area was assessed by linking patient identification codes of the Michigan Scleroderma Registry to the Metropolitan Detroit Cancer Surveillance System database. Patients were screened between the years 1973 and 2002, with additional followup to 2004. Standardized incidence ratios (SIRs) were calculated for selected malignancies (lung, liver, colon, breast, cervical, and prostate cancers, and non-Hodgkin's lymphomas), with stratification by sex and race.


Of 934 patients in the Scleroderma Registry, 538 were included in the study based on tri-county residency (436 females and 102 males). Of these, 45 first malignancies were noted (37 females and 8 males). Lung cancer (10 cases) was found to be the most common cancer in scleroderma patients. However, its incidence was not significantly different from that in the general population of metropolitan Detroit (SIR 1.23). Other types of cancer were examined, and no significant differences were found as compared with the rates in the local population, with 1 exception: black females with scleroderma had significantly higher rates of liver cancer (SIR 45.8).


Contrary to previous studies, this study did not find statistical evidence of an increased incidence of cancer in scleroderma patients, except for liver cancer. One possible reason is the high background rates of certain cancers in the metropolitan Detroit area. It may be necessary to consider local cancer rates when comparing different scleroderma cohorts.

Scleroderma (systemic sclerosis) is a chronic, multisystem, autoimmune disease that causes significant morbidity and mortality. Previous studies indicate a higher incidence of cancer in scleroderma patients (1–7). The strongest association has been with lung cancer (1–14). Isolated case reports suggest a higher incidence of bronchoalveolar type of lung carcinoma (15–17); however, none of the epidemiologic studies have confirmed that (1–14). Other types of cancers reportedly more common in scleroderma patients include breast cancer (1, 5, 6, 10, 18–21), nonmelanoma skin cancer (3), non-Hodgkin's lymphoma (1, 22, 23), esophageal cancer (24–26), and liver cancer (3).

It is difficult to precisely examine the possible increased risk of malignancy in scleroderma patients because of the rarity of this disease and therefore the small number of patients available for study. Although studies have suggested an overall higher risk of cancer in scleroderma patients, the standardized incidence ratios (SIRs) for individual cancers were inconsistent. Some studies showed a significant SIR for a particular type of cancer, whereas others did not (1–7). The absolute SIR values for the overall incidence of cancer and of individual cancers were also quite variable in the different studies.

In 2000, we reported the results of a population-based study of cancer in scleroderma patients, using data from the Michigan Scleroderma Registry and the Metropolitan Detroit Cancer Surveillance System (MDCSS) database (27). That study, which was conducted from 1973 to 1998 with followup to 2000, showed no overall increase in the incidence of cancer among the 490 scleroderma patients who were eligible for study. In addition, no increased risk of any of the specific types of cancer was noted.

Our present study is an extension of the original observations (27) and includes a larger scleroderma patient population and a longer followup with the MDCSS database. The objective of the present study was to determine the relationship (if any) between scleroderma and the occurrence of malignancy. If such a relationship exists, it can lead to better insight into the pathogenesis of these diseases and may modify surveillance methods in clinical practice. The lack of a relationship would provide reassurance to the patients and to their physicians.


  1. Top of page
  2. Abstract

In this study, scleroderma data were collected from January 1, 1973, to May 31, 2002, and followup for the occurrence of cancer was continued to May 20, 2004. The risk of cancer in scleroderma patients living in the Detroit metropolitan area was assessed by linking the following 2 local disease registries: the Michigan Scleroderma Registry and the MDCSS.

The Michigan Scleroderma Registry

The Michigan Scleroderma Registry is a population-based listing of scleroderma patients diagnosed and treated at local hospitals and clinics in the Detroit metropolitan area between the years 1954 and 2003 (28–30). The majority of patients (64%) live in the tri-county area of metropolitan Detroit, which consists of Wayne, Oakland, and Macomb counties. According to the 2000 US census data (31), metropolitan Detroit has an overall population of 3.9 million, of whom 68.0% are white, 25.8% are black, and 6.2% are of other races.

All scleroderma diagnoses were confirmed by review of relevant demographic, clinical, and laboratory data from the patients' medical records (31). If the diagnosis, eligibility, or other relevant clinical data were unclear, additional chart sources were sought and/or the patient's physician was contacted by telephone.

The diagnosis was recorded as definite if the patient satisfied the American College of Rheumatology (ACR; formerly, the American Rheumatism Association) 1980 criteria for the classification of scleroderma (32). The diagnosis was recorded as probable if the physician had diagnosed scleroderma and the patient had 3 or more features of limited cutaneous scleroderma (formerly called CREST syndrome, representing calcinosis, Raynaud's phenomenon, esophageal dysmotility, sclerodactyly, and telangiectasias) and did not fulfill criteria for another defined connective tissue disease, such as systemic lupus erythematosus (33). Scleroderma patients were identified by the International Classification of Diseases, Ninth Revision (ICD-9) code 710.1 (34).

The disease was designated as limited or diffuse based on skin involvement (as documented in the medical records), according to the method described by LeRoy et al (35). Limited disease was defined as skin thickening confined to the extremities and distal to the elbows and/or knees. Diffuse disease was defined as skin thickening involving the proximal extremities (proximal to the elbows and/or knees) or the trunk below the clavicles. If the medical record did not include a description of the extent of skin involvement but the above criteria were met, the case was designated as systemic scleroderma of unknown type. Cases of localized scleroderma (morphea and linear scleroderma) were excluded.

The date of diagnosis was defined as the date when scleroderma was first mentioned or the date when the first non–Raynaud's phenomenon symptom related to scleroderma was documented in the medical record (31). Patients were mailed an annual questionnaire to check vital status and disease status, with phone followup if the questionnaire was not returned (30). If the patient was not shown to be deceased according to the hospital discharge database or medical chart review, an annual search of the National Death Index was performed to determine whether the patient was deceased and, if so, the date of death (30, 31). The address listing was obtained from the medical record and/or from the commercial guide of household residences with dates of occupancy (Bresser's Guide, Detroit, MI) (31). Computerized phone directories and “reverse” directories were used to aid in maintaining correct address information (30).

In order to obtain as complete a census as possible, multiple sources were used for identification of cases: (a) cases recruited from the patient population of hospitals and outpatient clinics of Wayne State University/Detroit Medical Center, (b) the practices of local rheumatologists, 70% of whom participated in the study, (c) a computerized database of patients with the diagnosis of scleroderma (ICD-9 code 710.1 [34]) who were discharged from all of the 55 area hospitals, including records from the hospitals and outpatient clinics of the University of Michigan Medical Center, (d) the Southeast Michigan Chapter of the Scleroderma Foundation (a patient support group), and (e) other physician referrals and self referrals.

Although cases were collected from multiple sources, the census was probably not 100% complete. A counting correction technique, based on a capture–recapture method, (36) was used to estimate the total number of persons with scleroderma in the population. The analyses found the Scleroderma Registry to contain ∼85% of the possible predicted number of scleroderma cases (31).


The MDCSS (37, 38) is a population-based listing of cancer patients in the Detroit metropolitan area who were diagnosed between the years 1973 and 2004. It is a contributor to the Surveillance, Epidemiology, and End-Results (SEER) program of the National Cancer Institute, from which the cancer rates for the general US population are derived. The MDCSS is responsible for the registration of all patients who are newly diagnosed as having cancer among the residents of the Detroit metropolitan area. Followup on each patient is conducted annually to assess current vital status. Cancer registrars collect data from the medical records of area hospitals, selected neighboring hospitals, physicians' offices, clinics, radiation oncology facilities, private pathology laboratories, hospice facilities, nursing homes, death certificates, and the Michigan Office of Vital Statistics, resulting in an estimated capture rate of more than 99% of all cancer cases in the tri-county area (38).

Malignancies occurring in scleroderma patients during the study period (from 1973 to 2004) were identified from the MDCSS registry by linking patient identification codes. Patients from the scleroderma registry with a home address in the tri-county area at the time of scleroderma diagnosis were included in the calculation of SIRs. SIR calculations compare the observed number of cancers in the cohort with an expected number, which is obtained by applying the age-specific incidence rates from the general local population. Race-specific (black versus white) SIRs were calculated for all cancers and for each sex separately. Age groups consisted of 18 age clusters generally falling 5 years apart from age 0–4 years, 5–9 years, 10–14 years…75–79 years, 80–84 years, and 85+ years.

Age-, race-, and sex-specific cancer rates were determined from the metropolitan Detroit SEER registry data with census values from 1973 to 2001. Incidence rates for 2001 were used as surrogate incidence rates for 2002, 2003, and 2004 since at this time, the SEER registry is complete only through the year 2001. The point of inclusion in the study was the date of scleroderma diagnosis that occurred on or after January 1, 1973. The end point was the first of 3 possible dates: the date of cancer diagnosis, the date of death, or May 20, 2004.

SIR values were calculated using the program PAMCOMP, version 1.41 (39). PAMCOMP (person-years and mortality computation program) is an application for calculating person-years and SIRs using external files of cancer rates and patient information provided by the user. Also reported were person-years of followup, observed and expected numbers of cancers, and 95% confidence intervals (95% CIs) for the SIR values, calculated using Fisher's exact test (40) as given in the PAMCOMP software. The overall SIR calculation was made after grouping all invasive cancers together. For calculation of SIRs for individual sites of cancer, all cancers for which there were 3 or more cases were determined. Selected malignancies for which there were fewer than 3 cases were also measured, such as non-Hodgkin's lymphoma and colon cancer, which have a high prevalence in the general population.

Further analyses were also conducted on all data stratified by sex, race, and type of scleroderma. These analyses included age at scleroderma onset, clinical subgroups of scleroderma (i.e., limited and diffuse disease, and scleroderma sine sclerosis), serologic subgroups (i.e., presence or absence of antinuclear antibodies [ANAs], anti–Scl-70 antibodies, anticentromere antibodies [ACAs], and anti–U1-RNP antibodies), and specific histologic types of lung cancer.

Because this was a retrospective study and the records were collected from various physicians, a complete antibody profile was not available for all patients (31). Only 50.9% of the patients (274 of 538) had a complete antibody profile. We examined the distribution of antibodies in scleroderma patients as a function of scleroderma disease type. A previous study had suggested that based on demographic, clinical, laboratory, and serologic features and natural history, patients with scleroderma sine sclerosis should be included in the spectrum of limited scleroderma and should not be considered a separate disorder (41). Hence, for this analysis, the diffuse disease group was compared with the combined group of limited disease and scleroderma sine sclerosis. A predicted value for the number of people with cancer in each subgroup of patients was calculated using the general cancer rate in this study (number of cancers/total number of patients).

To test the assertion that a specific histologic type of lung cancer is more common in scleroderma patients, the histology codes for individual lung cancer cell types were tallied according to the nomenclature of the International Classification of Diseases for Oncology, Third Edition (ICD-O-3), 2000.

Approval for study of human subjects

The Institutional Review Board of Wayne State University approved this study. Compliance with HIPAA (Health Insurance Portability and Accountability Act) regulations was ensured.

Statistical analysis

Comparisons were examined by Fisher's exact test. Age at scleroderma onset was examined by t-test. P values less than or equal to 0.05 were considered significant.


  1. Top of page
  2. Abstract

At the time of this study, there were 934 unique patients in the Michigan Scleroderma Registry (Figure 1). A working subset of 538 patients with 45 primary cancers (8.4%) qualified for inclusion in the SIR calculations. To create the working subset, 396 of the 934 patients were excluded: 336 of them were not tri-county residents at the time of scleroderma diagnosis, 27 were diagnosed as having scleroderma before the 1973 start of the SEER rate calculations, and 33 developed cancer before the onset of scleroderma. Tri-county residency status was necessary to ensure capture by the MDCSS registry.

thumbnail image

Figure 1. Flow diagram showing the distribution of the 934 patients in the Michigan Scleroderma Registry.

Download figure to PowerPoint

Demographic data are shown in Table 1. The overall SIR for all invasive cancers in the total cohort was 0.91 (95% CI 0.66–1.22). There were 436 females with 37 cancers and 102 males with 8 cancers. The distribution of tumors between male and female patients was similar (P = 0.794). There were 355 white patients (66.0%) and 183 black patients (34.0%). The overall age at which patients were diagnosed as having scleroderma was 45.5 ± 15.5 years (mean ± SD). On average, blacks were diagnosed as having scleroderma 5.15 years earlier than whites (P < 0.001).

Table 1. Working cohort of scleroderma patients evaluated in this study
Race, sexNo. (%) of scleroderma patientsNo. (%) with cancers
White female288 (53.5)25 (55.6)
Black female148 (27.5)12 (26.7)
White male67 (12.5)7 (15.6)
Black male35 (6.5)1 (2.2)
Total538 (100)45 (100)

The frequencies and types of cancer found in the scleroderma patients are shown in Table 2. SIR calculations were performed on the following malignancies: lung, breast, prostate, cervix uteri, liver, non-Hodgkin's lymphoma, and colon. This group comprised 66.7% of all the cancers that occurred following a diagnosis of scleroderma. All cancers were invasive except for 2 in situ cervical cancers.

Table 2. Frequency and type of cancers in 45 scleroderma patients
Site of malignancyNo. of cancers
  • *

    Calculations of standardized incidence ratios were performed on the data for these malignancies.

  • Consists of 1 invasive and 2 in situ cancers.

Cervix uteri*3
Non-Hodgkin's lymphoma*2
Multiple myeloma2
Ill-defined sites2
Other female genital area1
Urinary bladder1

Lung cancer was the single most common type of cancer in this study. Of the 10 lung cancer cases (22.2%) reported here, 4 were adenocarcinomas, 2 were squamous cell cancers, 1 was a large cell cancer, 1 was a small cell cancer, 1 was an oat cell cancer, and 1 was a non–small cell cancer. While the MDCSS registry includes a unique code for bronchoalveolar carcinoma, this code was not assigned to any of these lung cancers.

The demographic breakdown of scleroderma disease states by sex, race, and disease type is shown in Table 3. Of the 538 scleroderma patients evaluated in this study, 187 (34.8%) had diffuse disease, 317 (58.9%) had limited disease, 9 (1.7%) had scleroderma sine sclerosis, and the nature of disease in 25 (4.6%) was unknown. Even though limited disease was the most common type of scleroderma in the total cohort, there was a significant difference in the distribution of disease types between black females and white females (P < 0.001). White females had significantly fewer cases of diffuse disease (n = 78) than was expected (n = 100) and more limited disease (n = 194) than was expected (n = 170). In contrast, black females had significantly more diffuse disease (n = 71) than was expected (n = 51) and less limited disease (n = 64) than was expected (n = 87). Of the 45 scleroderma patients with cancer, 10 (22.2%) had diffuse disease, 32 (71.1%) had limited disease, none had scleroderma sine sclerosis, and the nature of disease in 3 (6.7%) was unknown.

Table 3. Distribution of scleroderma disease states in all study patients, by sex and race*
Scleroderma disease typeWhiteBlackTotal
  • *

    Values are the number of scleroderma patients (number with cancer).

  • Value outside of the chi-square predicted value at P < 0.001.

Diffuse78 (4)23 (1)71 (5)15 (0)
Limited194 (19)42 (6)64 (6)17 (1)
Sine sclerosis6 (0)0 (0)2 (0)1 (0)
Unknown10 (2)2 (0)11 (1)2 (0)
Total288 (25)67 (7)148 (12)35 (1)

The demographic breakdown by race and sex of the scleroderma patients who had cancer is also shown in Table 3. The cancer incidence in patients with limited disease (32 malignancies) showed an odds ratio of 1.98 as compared with patients with diffuse disease (10 malignancies), which was significantly different (P = 0.042 by Fisher's exact test). We found no difference in the distribution of cancers in comparisons of male versus female patients (P = 0.581) and black versus white patients (P = 0.319) by Fisher's exact test.

The distribution of antibody types is presented in Table 4. The medical records showed incomplete reporting of antibody types. Reports of testing for Scl-70 antibodies were found in 323 of the 538 study patients (60.0%). In increasing order, U1-RNP antibody was tested in 338 patients (62.8%), ACA was tested in 440 patients (81.8%), and ANA was tested in 459 patients (85.3%). None of the stratified antibody groups were found to have cancer incidences that were statistically significantly different from those predicted by the general cancer rate found in this study (45 of 538 patients).

Table 4. Distribution of antibody types by scleroderma disease state, comparing the number of predicted and observed cancers*
Serologic statusNo cancerCancerP
  • *

    Values are the number of patients with scleroderma (predicted number of patients with cancer). Predicted numbers of patients with cancer were calculated from the global cancer rate; for example, for the number of antinuclear antibody (ANA)–positive scleroderma patients with cancer, the predicted number of patients with cancer was calculated as follows: 45/538 × (240 + 28) = 22.4. P values were determined by Fisher's exact test. ACA = anticentromere antibody.

ANA+24015228 (22.4)8 (13.4)0.034
ANA−18111 (1.6)1 (1.0)0.632
ACA+751010 (7.1)1 (0.9)0.632
ACA−17414319 (16.1)8 (12.6)0.086
Anti–Scl-70+39403 (3.5)4 (3.7)0.526
Anti–Scl-70−1319111 (11.9)4 (7.9)0.207
Anti–U1 RNP+16100 (1.3)0 (0.8)1.000
Anti–U1 RNP−16912412 (15.1)7 (11.0)0.414

SIR values for cancers are shown in Table 5. The cancers examined included lung, colon, liver, breast, cervical, and prostate cancers, and non-Hodgkin's lymphoma. There were a total of 45 cancers, 43 were invasive and 2 were in situ cervical cancers. SIR values that were significantly different from 1 are indicated in Table 5. Patients with scleroderma developed cancers (including all invasive cancers combined as well as individual cancer types) at a rate that was not statistically significantly different from that in the general population of metropolitan Detroit. This included lung cancer, which was the most common cancer seen in these patients. Only liver cancer showed a higher rate, about 7 times higher than expected in the general population. All of these liver cancers where found exclusively in females, making the odds more than 16 times greater than expected in the general female population.

Table 5. Standardized incidence ratios, stratified by sex and adjusted for age*
Cancer siteAll casesWhitesBlacks
Person- yearsNo. observedNo. expectedSIR (95% CI)Person- yearsNo. observedNo. expectedSIR (95% CI)Person- yearsNo. observedNo. expectedSIR (95% CI)
  • *

    Cancer of the cervix uteri represents invasive and in situ cervical cancers. SIR = standardized incidence ratio; 95% CI = 95% confidence interval; NHL = non-Hodgkin's lymphoma.

  • SIR significantly higher than 1.

Males and females            
 All invasive5,766.144347.350.91 (0.66–1.22)3,857.063134.770.89 (0.61–1.27)1,902.251212.840.93 (0.48–1.63)
 Lung5,895.29108.161.23 (0.59–2.25)3,971.7076.101.15 (0.46–2.37)1,924.1132.191.37 (0.28–4.00)
 Liver5,900.6030.417.35 (1.52–21.49)3,975.5200.241,925.0830.1816.32 (3.37–47.70)
 NHL5,887.9221.691.18 (0.14–4.28)3,962.3321.331.50 (0.18–5.43)1,925.5900.36
 All invasive4,838.103534.611.01 (0.70–1.41)3,264.772425.730.93 (0.60–1.39)1,573.42118.811.25 (0.62–2.23)
 Breast4,908.66911.170.81 (0.37–1.53)3,327.1588.440.95 (0.41–1.87)1,581.5112.730.37 (0.01–2.04)
 Lung4,951.4175.451.29 (0.52–2.65)3,364.0144.210.95 (0.26–2.43)1,587.4031.282.34 (0.48–6.83)
 Liver4,954.7730.1916.19 (3.24–47.32)3,366.4000.121,588.3730.0745.82 (9.44–133.91)
 NHL4,942.0021.241.61 (0.20–5.83)3,353.2120.992.02 (0.24–7.30)1,588.8800.25
 Cervix uteri4,929.8032.191.37 (0.28–4.01)3,347.7521.281.57 (0.19–5.66)1,582.0510.811.24 (0.03–6.92)
 All invasive928.0487.071.13 (0.49–2.23)592.3875.401.30 (0.52–2.67)335.6611.770.56 (0.01–3.14)
 Prostate935.7732.441.23 (0.25–3.59)600.1121.861.08 (0.13–3.89)335.6610.561.79 (0.05–9.95)
 Lung944.3931.292.33 (0.48–6.80)607.6831.003.00 (0.62–8.75)336.7100.33

Race-stratified and age-adjusted SIR values were then calculated for all cancers in all patients as well as separately in male and female patients (Table 5). There were 32 cases of cancer in the white patients, 31 were invasive and 1 was an in situ cervical cancer. White patients developed invasive cancers at about the same rate as the general white population (31 observed versus 35 expected). None of the individual cancers occurred in the white scleroderma patients at a rate that was statistically significantly different from that in the general white population of metropolitan Detroit. This included breast and lung cancer, which were the most common cancers reported. Fifteen lung cancers and 19 breast cancers would have been necessary to achieve statistical significance in the white population. Even though white males had an SIR value of 3 for lung cancer, this was not statistically significant; the calculated significant number would have been 4 lung cancers in the white male patients.

Table 5 also shows the SIR values for cancers in the black patients (n = 12). All cancers were invasive except for 1 in situ cervical cancer. Black patients with scleroderma developed invasive cancers (all cancers and individual cancers) at about the same rate as expected in the general black population of metropolitan Detroit. All cases of liver cancer (n = 3) occurred in the black female group, at a rate nearly 46 times greater than expected in the general black female population. Only 1 of the 35 black males with scleroderma developed cancer, which was not statistically significantly different from the expected value of 1.77.

Table 6 shows the SIR results for all cancers stratified by scleroderma disease types (diffuse and limited). From the original data set of 538 patients, 187 were diagnosed as having diffuse disease and 326 were diagnosed as having limited (n = 317) and sine disease (n = 9) combined. The disease type in 25 patients was unknown, including 3 patients who had cancer; all these data were excluded from the analysis. There was no further stratification by race, since the numbers of cancers were decreasing at each added level of stratification.

Table 6. Standardized incidence ratios, stratified by sex and adjusted for age*
Sex, cancer siteLimited scleroderma and scleroderma sine sclerosisDiffuse scleroderma
Person- yearsNo. observedNo. expectedSIR (95% CI)Person- yearsNo. observedNo. expectedSIR (95% CI)
  • *

    Cancer of the cervix uteri represents invasive and in situ cervical cancers. SIR = standardized incidence ratio; 95% CI = 95% confidence interval; NHL = non-Hodgkin's lymphoma.

  • SIR significantly higher than 1.

Males and females        
 All invasive3,661.793033.450.90 (0.81–1.28)1,790.261011.550.87 (0.42–1.59)
 Lung3,764.6375.871.19 (0.48–2.46)1,811.9621.891.06 (0.13–3.83)
 Liver3,768.5710.293.45 (0.09–19.23)1,812.4020.1020.23 (2.45–73.07)
 NHL3,760.0211.190.84 (0.02–4.68)1,808.2710.422.41 (0.06–13.42)
 All invasive3,060.432324.790.93 (0.59–1.39)1,495.7797.821.15 (0.53–2.18)
 Breast3,117.1468.010.75 (0.27–1.63)1,506.3522.540.79 (0.10–2.85)
 Lung3,149.6344.030.99 (0.27–2.54)1,515.2621.131.77 (0.21–6.41)
 Liver3,151.6210.147.38 (0.19–41.09)1,515.7020.0450.08 (6.06–180.80)
 NHL3,172.5610.901.12 (0.03–6.22)1,511.5710.273.64 (0.09–20.27)
 Cervix uteri3,130.4221.271.58 (0.19–5.69)1,551.9310.771.30 (0.03–7.24)
 All invasive601.3674.061.72 (0.69–3.55)294.4913.000.34 (0.01–1.87)
 Prostate609.0921.381.44 (0.17–5.22)294.4911.050.95 (0.02–5.29)
 Lung615.5130.754.01 (0.83–11.72)296.7000.54

In the limited scleroderma group, there were 30 invasive cancers and 2 in situ cancers. Patients with limited scleroderma developed invasive cancers at rates similar to those in the general population. Male patients with lung cancer were found exclusively in the limited scleroderma group. No cancer incidence was found to be higher than that expected in the general population. Even the incidence of liver cancers in female patients, which had an SIR of 7.38, also had wide confidence limits and was not statistically significant; only 1 cancer was observed, whereas 3 cancers would have been statistically significant. This was also true for lung cancer in male patients, which had an SIR of 4 and was not statistically significant; 3 cancers were observed, whereas 4 cancers would have been statistically significant.

In the diffuse scleroderma group, there were 10 invasive cancers. The occurrences of cancers were also similar to those expected in the general population. Only liver cancer was significantly higher (20-fold) than that in the general population of metropolitan Detroit. In the subgroup of females with diffuse disease, the liver cancer incidence was 50 times higher.


  1. Top of page
  2. Abstract

In the present study, a working subset of 538 patients with 45 primary cancers (8.4%) qualified for analysis. Women developed scleroderma at a rate ∼4 times higher than that in men, and the overall incidence of cancer was equivalent on a percentage basis by sex. Limited scleroderma was nearly twice as common as diffuse scleroderma. Black patients were diagnosed as having scleroderma ∼5 years earlier than whites. These results are consistent with the recently published data on the demographics of the Michigan Scleroderma Registry (31). Lung cancer and breast cancer were the 2 most common cancers seen in scleroderma patients, as well as in the general population. None of the SIR values for cancer were significant, except for the SIR for liver cancer, which occurred only in black females. However, there were only 3 cases of hepatoma, and the calculated SIR had a wide confidence interval, which calls into question the clinical significance of this finding. There was no special distribution of scleroderma disease types by either sex or race among patients who subsequently developed cancer.

One of the earliest studies to assess cancer risk in scleroderma was conducted by Duncan and Winkelmann (42). Of 2,141 scleroderma patients screened, 78 patients developed 87 cancers. The relative frequency of the types of cancers was similar to that in the general population. However, there was no mention of a control population, and no statistical analysis was described.

Since then, other investigators have done more analytical work. Abu-Shakra et al (2) reported a study based on 248 scleroderma patients. The SIR for all cancers was 2.1 times the overall rate in the local population (P < 0.0001). However, this was a chart review of office visits and, hence, not a population-based study.

A French study of 123 patients with systemic sclerosis who were observed over a mean period of 4 years showed 14 cancers (11.4%) (5). Lung cancer was the most common. All 3 patients with lung cancer had limited scleroderma and ACAs. In our study, there were 10 lung cancers: 7 occurred in patients with limited scleroderma, 2 in patients with diffuse scleroderma, and the disease type in 1 patient was unknown. Lung cancer also was the most frequent cancer in our cohort but was not significantly above background rates. Interestingly, none of our 7 patients with limited disease and lung cancer had ACAs. Only 1 patient with diffuse scleroderma and lung cancer had ACAs.

Rosenthal et al (3) reported their findings of a study based on 917 scleroderma patients who were hospitalized between the years 1965 and 1983. The data were obtained from the Swedish inpatient discharge registry, which covered 60% of the Swedish population in 1969 (the time when data collection for the study started), but the capture rate increased to ∼85% by 1983. There was a 1.5-fold increase in the overall risk of cancer. The SIR values were significant for lung cancer (SIR 4.9), nonmelanoma skin cancer (SIR 4.2), and primary liver cancer (SIR 3.3). However, the case ascertainment rate for scleroderma was incomplete and variable over the study period, whereas that for cancer was nearly complete. This could potentially bias the results, suggesting an apparent increase in the risk of cancer in scleroderma patients. In addition, data were obtained from hospital discharge records, and hence, this was not a population-based study. Similar to the findings of our study, there was a higher incidence of hepatoma.

Roumm and Medsger (4) reported the results of a retrospective review of 262 scleroderma patients over a 12-year period (1971–1982). There was a significant relative difference in malignancies between scleroderma cases and controls (1.81), which was predominantly due to an increase in cases of lung cancer. Although their control group was obtained from the same population, the time period was different (1969–1971).

Rothfield et al (8) have suggested that the prevalence of antibody to topoisomerase I (Scl-70) may be a risk factor for malignancy in scleroderma patients. Those investigators reported that cancer developed in 7 of 36 patients who had this antibody (19%), as compared with only 2 of 112 patients without the antibody (1.8%). Our findings do not confirm those results. We analyzed 323 scleroderma patients for an association between Scl-70 antibody and malignancy. Cancer developed in 7 of 86 patients with this antibody (8.1%), as compared with 15 of 237 patients without the antibody (6.3%) (data calculated from Table 4). This difference is not significant and supports the findings of a recently published study conducted by Derk et al (43).

More recently, Hill et al (7) reported the results of a population-based cohort study in South Australia. The SIR for all cancers was significantly increased (1.99). There was a significant increase in lung cancer (overall SIR 5.9) in both limited and diffuse scleroderma. Those authors reported that the South Australian Scleroderma Registry corresponds most closely to our Michigan Scleroderma Registry in its comprehensive means of ascertaining cases of scleroderma (27). Even though the capture rate was nearly 100% for cancer cases, possibly the rates of ascertainment of scleroderma cases were much lower, which could falsely increase the SIR values.

The growing consensus of studies reviewed in 2003 by Pearson and Silman (12) indicates that there is an increase in cancer risk related to scleroderma, especially lung cancer. It would provide a greater consistency of findings if our current findings supported the lung cancer risk, but they do not. The observed differences found in the other studies could be real or they could be accounted for by various combinations of the following factors. There could be differences in (a) the methodology of case ascertainment, (b) the degree of scrutiny applied to the case definition of scleroderma, (c) the degree of scrutiny applied to ensure completeness of case ascertainment, (d) the nature of the study (population-based versus practice-based or hospital-based, which introduces biases inherent in convenience sampling), (e) the genetic, racial, ethnic, behavioral, occupational, and environmental risk factors, and (f) the background incidence of cancer in different geographic areas.

To reiterate the last point, we need to remember that there are differences in the background incidence of cancer in different populations. For example, since 1975, Detroit has had the highest lung cancer rates of the 9 regional US collection centers that supply data to the SEER database. The mean ± SD age-adjusted rate of invasive lung cancer in the population of metropolitan Detroit from 1973 to 2000 was 76.4 ± 9.5 cases per 100,000 population. In contrast, the overall SEER rate for invasive lung cancer was 63.1 ± 5.7 cases per 100,000 (calculated from public access SEER rate tables) (44). The point to be made here is that when scleroderma databases are matched to cancer cohorts in defined geographic areas, it is necessary to consider the local cancer rates.

Our study has several strengths. It is a very large population-based study with a high degree of case ascertainment. It is the largest population of black patients with scleroderma ever studied for cancer risk assessment. Also, for the first time, we had an opportunity to perform a comparative study of cancer risk assessment between whites and blacks.

Some of the limitations of our study were as follows: we could not adjust for some known and unknown confounders (risk factors for cancer; e.g., dietary factors, cigarette smoking, and alcohol use). A healthier lifestyle among scleroderma patients (e.g., fewer smokers) may reduce the risk of cancer. Also, the genetic predisposition (if any) may be different between whites and blacks. The SIR for liver cancer was high. Although none of the 3 patients with hepatoma were known to have primary biliary cirrhosis (a risk factor for liver cancer), no data on alcohol use or hepatitis B or hepatitis C status were available for these patients. We also did not have data on noncervical in situ cancers and nonmelanoma skin cancers. Data on serology and type of disease were incomplete. Moreover, we had no information on tumor behavior (aggressiveness of cancer), response rate to standard therapy, and natural history of cancer in scleroderma patients.

We conclude that contrary to previous studies, the present study did not show evidence of a significantly increased incidence of cancer in patients with scleroderma. Among individual cancers, only the liver cancer rate was increased, although the significance of that finding was questionable because of the small number of cases and the wide confidence interval. The precise matching, as well as the population-based nature of the case cohort, may account for our findings and supports their validity.


  1. Top of page
  2. Abstract