ACR Boston 2014 | Daily Highlights

Development and External Validation of a Five-Year Mortality Risk Stratification Tool for Early Diffuse Systemic Sclerosis Patients

Abstract: 2997
Presenter: Robyn T. Domsic
Co-Authors: Svetlana I. Nihtyanova, Mary Lucas, Stephen R. Wisniewski, Michael J. Fine, C. Kent Kwoh, Christopher P. Denton and Thomas A. Medsger Jr.

Background:
Knowledge of mortality risk and predictors is important in systemic sclerosis (SSc) patient care and clinical trial design. There is no validated 5-year mortality model in early diffuse SSc (dcSSc); a high risk population. The objective of this study was to derive and externally validate a 5-year mortality risk stratification tool in early dcSSc patients.

Methods:
The derivation cohort was a prospectively enrolled inception cohort of adult early dcSSc patients first seen between 1980 and 2009 and enrolled in a US Scleroderma Center databank. Early dcSSc was defined as < 2 years from the first SSc symptom and proximal skin thickening. Predefined candidate predictor variables at the first visit (demographic, history, exam, lab values, organ system objective tests) were placed into a stepwise multivariable logistic regression model. Regression diagnostics were performed. Beta-estimates were rounded to the nearest 1, and then summed for a total score, which was then classified into low, moderate and high risk morality categories. The external validation cohort was a large UK Scleroderma Center databank, with patients meeting identical inclusion criteria and an initial visit between 2000 and 2008. Area under the curve (AUC) was calculated to assess discrimination in the derivation and validation cohorts, and stratum-specific chi-square analysis performed.

Results:
In the US derivation cohort, 760 early dcSSc patients were identified, of whom 388 were Caucasian and had all objective testing required to assess internal organ involvement at the first visit. The mean age was 50.4±13.3 years, 76% female and median disease duration 0.93 years (IQR 0.63, 1.33). The mean skin thickness score was 26.1 ± 11.7. After 5 years, 110 of the 388 (28.4%) had died. 144 patients composed the UK validation cohort, with no significant differences in age, gender or skin score from the US cohort. Median disease duration was longer at 1.02 (0.78,1.45) years (p=0.03). At 5 years 38/144 (26.4%) had died, which was not different (p=0.65) from the US cohort.

Significant independent predictors of 5-year mortality at first visit were age, gender, tendon friction rubs, GI involvement, RNA polymerase III antibody and anemia (Table 1). The AUC in the US derivation cohort was 0.79 (95% CI 0.74-0.84) for the overall model and 0.74 (0.65-0.82) in the UK cohort. Using the total sum score, risk stratification was defined as low (≤ 0), moderate (1-2) and high (≥3) risk. The AUC for the 3-level risk stratification in the US derivation cohort was 0.76 (0.72-0.81) compared to 0.68 (0.60-0.76) in the UK validation cohort (p=0.24). There were no significant differences in mortality rate between each strata of the risk stratification model in the US and UK cohorts.

Conclusions:
We have developed and externally validated an easy-to-use 3-level risk stratification tool for 5-year mortality in early dcSSc patients. This tool requires only history, exam and bloodwork.

Disclosure of Interest:
R. T. Domsic, None; S. I. Nihtyanova, None; M. Lucas, None; S. R. Wisniewski, None; M. J. Fine, None; C. K. Kwoh, None; C. P. Denton, None; T. A. Medsger Jr., None.

Comment:
The recently published ASTIS trial could show significantly improved survival after more than two years follow up in patients undergoing autologous hematopoietic stem cell transplantation – at the cost of high treatment related mortality of 10% in the first year. Thus, this treatment approach might be an option for patients at highest risks for death due to SSc. However, information is limited on reliable and feasible mortality prediction models in SSc. This methodologically very well performed study offers an easy to use risk stratification tool for patients with SSc. Further studies have to show whether this tool is also useful to stratify patients for certain therapies such as autologous hematopoietic stem cell transplantation and how it performs compared to other recently developed prediction models.

---

> zurück zur Übersicht