Predictors and outcomes of scleroderma renal crisis: The high-dose versus low-dose D-penicillamine in early diffuse systemic sclerosis trial

Abstract

Objective

The reported frequency of scleroderma M01-R renal crisis (SRC) in diffuse systemic sclerosis (SSc; scleroderma) is 15–20%. Early use of angiotensin-converting enzyme (ACE) inhibitors has markedly improved outcome. The present analysis reexamines the prognostic factors for and outcome of SRC in a prospective cohort of patients with early diffuse SSc.

Methods

We retrospectively evaluated the cohort of SSc patients who participated in the High-Dose Versus Low-Dose D-Penicillamine in Early Diffuse SSc trial. Patients with diffuse cutaneous scleroderma were enrolled if their disease duration was <18 months. Because the trial failed to show a difference between treatment groups, the data were pooled.

Results

One hundred thirty-four SSc patients entered the observation period a mean ± SD of 0.8 ± 0.3 years after onset of SSc. SRC occurred in 18 patients a mean ± SD of 0.9 ± 1.1 years after entry. During a mean ± SD 4.0 ± 1.1 years of followup after entry, 9 of the 18 patients died (mean ± SD 0.6 ± 0.9 years after SRC onset). Baseline characteristics that predicted SRC included a modified Rodnan skin thickness score of ≥20 (P < 0.01), enlarged cardiac silhouette on radiograph (P = 0.04), large joint contractures (wrist, elbow, knee) (P = 0.008), and prednisone use at entry (P = 0.01). Baseline characteristics that did not predict SRC included age, sex, race, Health Assessment Questionnaire score, fist closure, handspread, lung involvement, muscle weakness, erythrocyte sedimentation rate, and platelet count. In 5 of 10 subjects for whom at least 2 sequential skin scores were available, skin scores increased significantly (P = 0.012) in the 6 months before onset of SRC.

Conclusion

SRC occurred in 13% of patients soon (mean 11 months) after entry into the cohort. Predictors of SRC identified in this study included higher than average skin score, prednisone use at study entry, large joint contractures, and heart enlargement. Our data suggest, however, that low-dose prednisone alone was not associated with the onset of SRC, except in the appropriate clinical setting. Although ACE inhibitors and dialysis are now readily available, SRC continues to be associated with poor survival (in this study, 50% of patients with SRC died).

Prior to the introduction of angiotensin-converting enzyme (ACE) inhibitors in 1980, scleroderma renal crisis (SRC) was the leading cause of death in patients with systemic sclerosis (SSc) (1). Since 1984, Steen et al have reported their experience with SSc renal disease, derived from an observational study in which a large group of SSc patients were followed up prospectively since 1972 (2–4). They reported that 1-year survival in SRC patients treated with ACE inhibitors was 76%, compared with 18% in SRC patients not treated with ACE inhibitors, and that 5-year survival in SRC patients treated with ACE inhibitors was 60%.

Risk factors that Steen et al and other investigators have identified as being predictive of SRC include new anemia, new cardiac events (e.g., pericardial effusion or congestive heart failure), use of prednisone in dosages >15–20 mg/day, diffuse skin thickening, rapidly progressive skin thickening, SSc disease duration <4 years, and the presence of anti–RNA polymerase III antibody (1–5). Urinary abnormalities and elevations of serum creatinine or plasma renin preceding the onset of SRC have not been shown to predict SRC.

More recently, a cohort of 134 patients with diffuse SSc of recent onset (<18 months) were followed up prospectively for a mean of 4 years, as participants in the High-Dose Versus Low-Dose D-Penicillamine in Early Diffuse SSc (D-Pen) trial (6). Because the 3 primary outcomes in the high-dose and low-dose penicillamine groups were not significantly different, the data for all patients were combined. Over the 4 years of followup, 18 patients developed SRC. Because data at the time of entry and during 4 years of followup were accumulated regularly (according to protocol), this group of patients provides a unique opportunity to reevaluate predictors of SRC and its outcomes in the context of a prospective cohort design. Here, we contrast the baseline characteristics of SSc patients in whom SRC developed with the baseline characteristics of SSc patients in whom SRC did not develop, in order to examine the factors that predicted SRC and to document survival in the patients with SRC.

PATIENTS AND METHODS

Patients.

In the D-Pen trial, 17 centers enrolled 143 patients with SSc who met the American College of Rheumatology (formerly, the American Rheumatism Association) criteria for the classification of SSc (7) and who had diffuse cutaneous scleroderma (sclerodermatous induration of skin proximal as well as distal to the elbows and knees, with or without facial involvement) of <18 months duration since the onset of the first SSc manifestation other than Raynaud's phenomenon (6). Only the 134 patients who took ≥1 dose of study medication are included in this analysis (9 patients signed the consent forms but never took a dose of study medication). The primary outcomes of the trial were the course of skin thickening over 2 years, the frequency of new-onset SRC, and mortality over 4 years. Data were gathered prospectively to evaluate these end points. The trial was conducted from January 1991 through December 1996, but followup data were collected through 1997. All patients signed a consent form approved by the institutional review board at their centers.

Corticosteroids, if taken, had to have been at a stable dosage of ≤10 mg of prednisone (or equivalent) per day for at least 1 month before study entry. During the trial, patients could receive up to 6 bursts of prednisone in doses ≤15 mg daily (although such bursts were rarely administered). Treatments with putative disease-modifying properties had to have been discontinued for at least 1 month before study entry, as previously reported (6). Other exclusions were also previously reported (6).

Diagnostic tests.

Skin thickening was assessed at baseline and every 3 months for the first 2 years, using the modified Rodnan skin thickness score technique (skin score) (5, 6). At baseline and every 6 months during the first 2 years, assessments included active handspread, fist closure, oral aperture, Health Assessment Questionnaire Disability Index (HAQ-DI), serum creatine kinase (CK), manual muscle testing by physical examination, joint tenderness and swelling, presence and number of contractures of wrists, elbows, and knees (large joint contractures), presence and number of palpable tendon friction rubs, and routine chemistries (6).

At baseline and every 12 months, pulmonary function tests (forced vital capacity [FVC], forced expiratory volume in 1 second, total lung capacity, and diffusing capacity for carbon monoxide [DLCO]) were performed, and a chest radiograph (measuring heart size and the presence of interstitial change) and physician assessment were used to evaluate heart and lung involvement. Renal involvement was evaluated by a 24-hour urine test for creatinine clearance and protein content.

SRC was determined to be present when the patient's physician–investigator detected renal insufficiency (serum creatinine ≥2.0 mg/day, or a doubling of serum creatinine above the value at baseline, in the absence of another defined cause) and/or malignant hypertension (systolic blood pressure [BP] ≥160 mm Hg or diastolic BP ≥110 mm Hg on at least 2 occasions, a minimum of 12 hours apart), accompanied by persistent urine abnormalities or evidence of microangiopathic hemolytic anemia (MAHA). The following equation may help to explain the definition of SRC: SRC = (↑ serum creatinine) and/or (↑ BP + abnormal urinalysis or MAHA).

Although SRC was treated at individual centers using medical standard of care, including ACE inhibitors and dialysis as necessary, many of the SRC patients were treated at community medical centers other than those involved in the D-Pen trial. Although we tried to retrieve as much information as possible about the events leading up to SRC and its treatment at these outside medical facilities, the information we did obtain was often incomplete.

Lung involvement was considered present if the DLCO was ≤70% of predicted, FVC was ≤75% of predicted, or definite interstitial changes were noted on the chest radiograph. Renal involvement was defined by a serum creatinine level that was higher than the upper limit of normal, a 24-hour endogenous creatinine clearance of <70 ml/minute (corrected to 1.73 mm³ body surface area), or the occurrence of SRC. Muscle involvement was defined by a CK level that was ≥200% of the upper limit of normal or by a muscle strength score ≤4 (based on a 5-point Likert scale) in the proximal muscles (shoulder girdle or hip girdle). Joint involvement was considered present if the tender joint count was ≥1. Heart involvement was defined by a history or presence of congestive heart failure, cardiac arrhythmia requiring medication, pericarditis, moderate-to-large pericardial effusion, or cardiomegaly (cardiothoracic ratio >0.5 on chest radiograph).

Followup data on patients participating in the study were derived from questionnaires or in-person examinations, which allowed documentation of essential data for 99% of patients, for a mean ± SD of 4.0 ± 1.1 years. Review of the case report forms documented the type, dosage range, total amount (in mg), and duration of corticosteroid therapy before study entry and before the onset of SRC.

Statistical analysis.

The statistical analyses were performed using Stata (College Station, TX) and SAS (Cary, NC) software. All continuous data are displayed as the mean ± SD, unless otherwise specified. Group means were compared using Student's 2-tailed, unpaired t-test, and dichotomous variables were compared by use of Fisher's exact test. Entry data for patients in whom SRC eventually developed were compared with such data for patients in whom SRC did not develop. For logistic regression, clinical subsets were developed based on dichotomous variables (divisions occurring naturally or when continuous variables were made dichotomous by division above or below a cutoff point, usually the median).

Crude odds ratios (ORs) and 95% confidence intervals (95% CIs) were estimated using logistic regression, with SRC as a dichotomous outcome variable and the explanatory variables as dichotomous variables. Only explanatory variables with P values less than 0.05 by chi-square testing during univariate logistic regression were selected for further multivariate logistic regression. SRC, as the outcome variable of interest, and independent variables of interest were entered as dichotomous explanatory variables into backward, stepwise multivariate logistic models. Adjusted ORs and 95% CIs were used to model the associations of each predictor, with SRC as the outcome. The R² value was the measure of how well the logistic regression model explained the outcome (8). The effect of explanatory variables on SRC was also analyzed by life-table analysis (Kaplan-Meier method).

RESULTS

The details of the randomized, controlled trial of high-dose (1,000 mg daily) versus low-dose (125 mg every other day) penicillamine for the diffuse cutaneous scleroderma variant of SSc have been published elsewhere (6). In summary, the course of the skin score and the frequencies of SRC and mortality in the low-dose and high-dose groups were not dissimilar. Skin thickness scores improved to similar degrees in the 68 patients (pooled from both treatment groups) who completed 2 years of drug treatment. Data for all 134 patients who entered the trial were pooled for the analysis of the frequency of SRC and mortality. Because 1 patient was lost to followup, only 133 patients were analyzed for SRC. Mortality and the incidence of SRC were not different in the 66 high-dose patients and the 68 low-dose patients, who were followed up for a mean ± SD of 4.0 ± 1.1 years.

Patients entered the trial a mean ± SD of 0.8 ± 0.3 years after the onset of SSc (dated from the first manifestation of SSc other than Raynaud's phenomenon). During a mean ± SD followup period of 4.0 ± 1.1 years following entry into the study, 18 patients (13% of the cohort) developed renal crisis, a mean ± SD of 0.9 ± 1.1 years after study entry (1.7 ± 1.1 years after the onset of SSc). In half of these patients, SRC occurred within 0.4 years after entry. Nine (50%) of the 18 patients died, a mean ± SD of 0.6 ± 0.9 years after SRC occurred.

Baseline characteristics of the 18 patients in whom SRC developed were compared with those of the 115 patients in whom SRC did not develop (see Table 1 for continuous variables and Table 2 for dichotomous variables). We observed that at baseline, the modified Rodnan skin score and systolic BP were higher, the oral aperture and hematocrit (in females) were lower, and heart involvement, contractures of large joints, and prednisone use were more frequent in patients who developed SRC than in patients who did not develop SRC (P = 0.005–0.042).

Table 1. Baseline continuous parameters/variables of the 134 study participants*

Variable	SRC (n = 18)	Non-SRC (n = 116)	P
* Values are the mean ± SD. SRC = scleroderma renal crisis; HAQ = Health Assessment Questionnaire; DLCO = diffusing capacity for carbon monoxide. † P < 0.05.
Demographic
Age, years	44.9 ± 12.4	43.7 ± 12.4	0.705
History
Duration of Raynaud's at entry, days	276 ± 178	627 ± 1,310	0.304
Physician examination
Systolic blood pressure	130.3 ± 23.8	118.7 ± 10.3	0.018†
Diastolic blood pressure	76.3 ± 11.1	72.6 ± 10.3	0.169
Maximum oral aperture, mm	41 ± 8	46 ± 10	0.042†
Handspread, right, mm	178 ± 23	174 ± 29	0.634
Fist closure, right, mm	24 ± 16	24 ± 22	0.804
Skin score (0–51)	25.9 ± 6.2	20.5 ± 8.0	0.010†
Tender joint count (0–8)	1.3 ± 2.7	1.5 ± 2.3	0.654
Swollen joint count (0–8)	1.3 ± 2.3	0.8 ± 1.4	0.172
HAQ Disability Index (0–3)	1.21 ± 0.60	1.01 ± 0.68	0.238
Large joint contractures (0–6)	2.6 ± 1.6	1.4 ± 1.7	0.005†
Tendon rubs (0–10)	1.3 ± 1.4	0.7 ± 1.2	0.607
Laboratory
Hematocrit, male, %	40.0 ± 2.7	41.5 ± 3.1	0.425
Hematocrit, female, %	36.1 ± 4.2	38.6 ± 3.4	0.014†
White blood cell count, × 1,000/mm^3	7.8 ± 2.2	8.4 ± 2,3	0.321
Platelet count, × 1,000/mm^3	377 ± 113	336 ± 102	0.130
Erythrocyte sedimentation rate, mm/hour	29 ± 17	23 ± 17	0.200
Creatine kinase, % of upper limit of normal	79 ± 84	83 ± 131	0.909
DLCO, % predicted	73.1 ± 17.2	76.0 ± 18.2	0.519
Forced vital capacity, % predicted	83.6 ± 11.9	83.1 ± 17.8	0.910
Creatinine clearance, ml/minute	90.5 ± 29.8	93.2 ± 31.1	0.732
Serum creatinine, mg/dl	0.86 ± 0.02	0.89 ± 0.19	0.471

Table 2. Baseline dichotomous parameters/variables of the 134 study participants*

Variable	SRC (n = 18)	Non-SRC (n = 116)	P
* Values are the no. (%). Dichotomous variables are defined within the text. SRC = scleroderma renal crisis; HAQ-DI = Health Assessment Questionnaire Disability Index; ACE = angiotensin-converting enzyme. † P < 0.05.
Sex
Female	15 (83)	89 (77)	0.547
Male	3 (17)	27 (23)
D-penicillamine dose
High	8 (44)	58 (50)	0.687
Low	10 (56)	58 (50)
Prednisone use	10 (56)	30 (26)	0.010†
Digital tip ulcer	2 (11)	12 (10)	0.950
Non–digital tip ulcer	2 (11)	16 (14)	0.540
Proximal muscle weakness	4 (22)	14 (12)	0.088
Muscle involvement	5 (28)	17 (15)	0.199
Tendon friction rub	10 (56)	39 (34)	0.087
Large joint contractures	17 (94)	58 (50)	0.001†
Joint involvement	4 (22)	45 (39)	0.152
Renal involvement	7 (39)	30 (26)	0.280
Heart involvement	7 (39)	20 (17)	0.041†
Lung involvement	11 (61)	59 (51)	0.505
HAQ-DI score ≥1.0	13 (72)	56 (48)	0.074
Age ≥45 years	10 (56)	57 (49)	0.687
Skin score ≥20	16 (89)	50 (43)	0.000†
Medication
ACE inhibitor	2 (11)	9 (8)
Calcium channel blocker	5 (28)	43 (37)

Information about prednisone use before entry into the study was available for 133 patients. Forty patients were taking prednisone at entry (average dose 7.4 mg daily of prednisone equivalent). Ten (56%) of the 18 patients who eventually developed SRC were taking prednisone at study entry, compared with only 30 (26%) of the 115 non-SRC patients (P = 0.01). The mean ± SD duration of prednisone therapy and the mean ± SD daily prednisone dosage prior to entry in those taking prednisone were 46 ± 45 days and 9.7 ± 10.8 mg, respectively, in patients who eventually developed SRC compared with 134 ± 180 days and 6.0 ± 5.4 mg, respectively, in patients who did not develop SRC (P not significant [NS]). In the 10 SRC patients who were taking prednisone at study entry, SRC occurred a mean ± SD of 0.6 ± 0.7 years after entry, compared with 1.3 ± 1.4 years in the 8 SRC patients who were not taking prednisone at entry (P NS). In 9 of the 10 SRC patients who were taking prednisone, the dosage remained ≤10 mg daily until the time of SRC onset. The cumulative prednisone dose from the time of entry until the onset of SRC ranged from 350 mg to 6,690 mg. The tenth SRC patient, however, was treated for myositis with prednisone, in dosages up to 60 mg daily prior to the onset of SRC.

Using univariate logistic regression and life-table methods, 4 potential baseline predictors of SRC were identified: skin score ≥20, large joint contractures, prednisone use, and enlarged heart shadow (heart involvement) (Table 3). When these 4 variables were entered into stepwise, backward multivariate logistic regression, only skin score ≥20 (multivariate OR 6.68, 95% CI 1.41–31.70) and large joint contractures (multivariate OR 9.54, 95% CI 1.18–77.24) remained as independent contributors to predicting SRC (R² = 0.1603). The addition of other explanatory variables did not appreciably increase the R² value.

Table 3. Baseline variables that correlated with or predicted scleroderma renal crisis (n = 133)*

Variable	Logistic Regression				Kaplan-Meier
	OR	95% CI	P	Group at risk	Log-rank P	Wilcoxon P
* OR = odds ratio; 95% CI = 95% confidence interval.
Skin thickness	10.00	2.213–45.907	0.003	Skin score ≥20	0.0004	0.0007
Large joint contractures	16.12	2.075–125.254	0.008	Presence of large joint contractures	0.0006	0.0007
Prednisone use	3.63	1.304–10.051	0.014	Prednisone use	0.0201	0.0202
Heart involvement	2.93	1.010–8.482	0.048	Heart involvement	0.0104	0.0075

Prednisone use has been reported to be an important predictor of SRC (4). Therefore, we explored a potential interaction of prednisone use with high skin score (≥20) and large joint contractures, which, as noted above, were the best predictors of SRC (Figure 1 and Table 4). When neither a high skin score nor large joint contractures were present, there was no apparent risk of SRC, with or without prednisone use. Patients with either a high skin score or large joint contractures had a mild (but not significant) increase in SRC risk, with or without prednisone. In patients with both a high skin score and large joint contractures, however, the risk of SRC was significantly increased compared with that in patients who had neither a high skin score nor large joint contractures, regardless of whether they were (43% versus 0% [P < 0.033]) or were not (21% versus 0% [P < 0.008]) taking prednisone. These data suggest that prednisone use alone was not associated with an increased risk of SRC; the presence of high skin scores and contractures together, however, increased the risk of SRC, particularly in the presence of prednisone (21% in patients not taking prednisone and 43% in patients taking prednisone).

Figure 1. Risk for development of scleroderma renal crisis (SRC) in patients with systemic sclerosis. The risk increases with a skin score ≥20 and/or large joint contractures at baseline, particularly in patients taking prednisone. ∗ = P < 0.33–0.008 in patients with a skin score ≥20 and large joint contractures versus patients with neither characteristic.

Table 4. Risk of developing SRC based on skin score, presence of large joint contractures, and prednisone use at baseline*

Characteristic	Prednisone use at baseline
	No (n = 93)	Yes (n = 40)
* Values are the no. (%). SRC = scleroderma renal crisis. Risk is based on a mean ± SD followup of 4.0 ± 1.1 years. † P = 0.008 versus neither skin score ≥20 nor contractures, by Fisher's exact test. ‡ P = 0.033 versus neither skin score ≥20 nor contractures, by Fisher's exact test.
Neither skin score ≥20 nor large joint contractures	0/32 (0)	0/9 (0)
Only skin score ≥20	1/14 (7)	0/3 (0)
Only large joint contractures	1/19 (5)	1/7 (14)
Both skin score ≥20 and large joint contractures	6/28 (21)†	9/21 (43)‡

Steen et al reported that rapid progression of skin thickening frequently occurred just before onset of SRC (and is a predictor of SRC) (2). Of our 18 SRC patients, only 10 had had 2 or more sequential 3-monthly skin scores in the 6 months before onset of SRC. Of these 10 patients, 5 had documented increases in skin score (mean ± SD increase 11.6 ± 5.1 units) in the 6 months before onset of SRC.

DISCUSSION

When interpreting these data, it is important to recognize that the study from which these patients were taken (the D-Pen trial) was not designed to predict scleroderma renal crisis but only to measure and document its occurrence in response to medication (6). Nevertheless, the data presented here were gathered in a careful, rigorous, prospective manner and merit close analysis. It is true that there may be some center bias in these data, because the frequency of SRC was appreciably higher in one northeastern center. At that center, skin scores, prednisone use, HAQ-DI scores, and frequency of contractures in their 7 SRC patients were equivalent to the values for the SRC group as a whole (n = 18). Despite this potential bias, the overall frequency of SRC (13%) is similar to that reported by Steen et al (1, 2). The 50% mortality in our SRC cohort, which occurred a mean ± SD of 0.6 ± 0.9 years after SRC onset, is impressive although discouraging, because good medical care (including ACE inhibitors and dialysis) was available to our SSc patients at study centers and/or in the local community. Our mortality rate from SRC (50% at 0.6 years) was higher than that reported by Steen et al (1-year mortality rate from SRC only 24% in patients taking ACE inhibitors). In the present series, all but 1 of the 18 patients with SRC were treated with ACE inhibitors.

Our experience contrasts with that reported by Giersson et al, who observed SRC to be an infrequent cause of death in a cohort of 100 SSc patients who entered the study a mean of 5 years after onset of SSc (9). We suggest that in our study, the short duration of SSc at the time of entry and the fact that all of the patients had diffuse SSc are the important differences between the studies. Steen et al reported that 75% of the occurrences of SRC occurred within 4 years of SSc onset, primarily in patients with diffuse cutaneous scleroderma (1). In the present study, all of the patients had diffuse SSc, and all were enrolled within 18 months of SSc onset, a scenario likely to be associated with a higher rate of SRC. These observations suggest that if the goal of a study is to impact the onset and/or course of SRC, patients with diffuse SSc must be recruited very early in the course of SSc.

Because all of our patients had diffuse scleroderma, we could not address the issue of whether diffuse scleroderma per se was a predictor of SRC, but that is already an accepted fact. Our data indicated that a high modified Rodnan skin thickness score (≥20) and the presence of large joint contractures were very strong independent predictors of SRC. This supplements previous observational data reported by Steen et al suggesting that diffuse skin thickening is a predictor of SRC (2). In addition, 5 of the 10 SRC patients who had had at least 2 skin scores obtained in the 6 months before onset of SRC were documented to have had an increase in skin thickness scores before development of SRC (mean increase 11.5 units). These data support the observations of Steen et al that rapid progression of skin thickening in the months before SRC predicts SRC (2). Like skin thickening, large joint contractures are undoubtedly the result of the fibrosing process, which also affects joints and tenosynovium, and for that reason their presence connoted disease severity and predicted SRC. This suggests that more severe skin fibrosis, as assessed in several different ways (i.e., high skin score, large joint contractures, progressive skin thickening), predicted SRC.

Although we included clinical cardiac manifestations (i.e., pericardial effusion, arrhythmia requiring treatment, and congestive heart failure) in the definition of heart involvement, these were infrequent signs in our study. The most frequent indicator of heart involvement in the present study was simply an enlarged heart shadow on chest radiography. Although the present data showed a correlation of an enlarged heart silhouette with SRC, the study design using only chest radiography did not allow any further speculation regarding the cause of enlarged cardiac silhouette. This supplements data presented by Steen et al, who showed that the presence of symptomatic pericarditis, pericardial effusion, or congestive heart failure put SSc patients at higher risk of SRC (2).

The concern that corticosteroid use may exacerbate the clinical manifestations of SSc has been long-standing (10). A recent case–control study retrospectively reviewed the relationship of SRC to corticosteroid use in the 6 months preceding the onset of SRC (4). That study suggested that new use of prednisone in dosages ≥15 mg daily was statistically significantly associated with the subsequent onset of SRC, whereas new use of low-dose steroids (≤15 mg daily) or continuous use of steroids for >6 months was not associated with increased risk of SRC. In contrast, our present data suggest that even lower dosages of prednisone (mean 7.4 mg daily) were associated with the onset of SRC, and then only in the appropriate clinical setting (i.e., skin score ≥20 and large joint contractures).

We further explored interactions of prednisone with high skin scores and large joint contractures (both measures of skin involvement). In the absence of prednisone use, the risk of SRC in patients with both a skin score ≥20 and large joint contractures rose to 21%; however, when prednisone was added to the combination of high skin score and large joint contractures, the risk of SRC increased to 43%. These data suggest that restraint and caution must be used when prescribing corticosteroids to patients with SSc, especially when considering prednisone use in those with early diffuse disease, high skin scores, and large joint contractures.

With the frequency of SRC being 13% over 4 years, it should be noted that future studies using SRC as an outcome will need to plan accordingly to take into account the length of observation that will be needed. The same can be said for studies that use mortality as an outcome.

In conclusion, SRC occurred in 13% of 133 patients with diffuse SSc within a mean of 11 months of entry into the cohort (or within a mean of 20 months after the onset of SSc). In spite of the availability of ACE inhibitors, 50% of those patients died. Predictors of SRC were higher than average skin score (≥20), contractures of large joints (wrists, elbows, and/or knees), enlarged cardiac silhouette on chest radiograph, and, in the context of a skin score ≥20 and large joint contractures, prednisone use. Corticosteroids, even at low dosages, should be used with great caution in the setting of early diffuse SSc.