Background/Purpose: Identifying systemic sclerosis (SSc) subsets that predict mortality and morbidity could provide useful prognostic information. We undertook this study to compare the predictive ability of different approaches to subsetting SSc.

Methods: SSc subjects from the Canadian Scleroderma Research Group cohort were studied. Three approaches to subsetting were used: Leroy subsets based on skin involvement (limited (lcSSc) and diffuse cutaneous (dcSSc) subsets), serological subsets (anti-centromere (ACA), anti-topoisomerase I (ATA) and anti-RNA polymerase III (RNAP)), and unsupervised cluster analysis based on the items in the ACR/EULAR 2013 classification criteria, of which 3 clusters were identified (cluster 1 (ACA negative subjects with digital ulcers (DU)), cluster 2 (ACA positive subjects), and cluster 3 (ACA negative subjects with no DU)). Morbidity was defined as forced vital capacity (FVC) < 70% predicted, interstitial lung disease (ILD), pulmonary hypertension (PH) and impaired health-related quality of life (defined as SF-36 Physical Component Summary (PCS) score < 40). Kaplan Meier curves were generated to compare the time to event between the various subsets. Log rank p values < 0.05 were considered statistically significant.

Results: 805 subjects were included (86.1% (N=693) female, 49.8% (N=401) dcSSc, disease duration since onset of first non-Raynaud’s disease manifestation 10.8+9.2 years). Subsetting based on autoantibodies and unsupervised clustering (but not Leroy classification) predicted mortality, with ACA having better survival than RNAP and cluster 1 having better survival than the 2 other clusters. All three approaches to subsetting predicted FVC < 70% and development of ILD: dcSSc was worse than lcSSc, ACA was better than ATA and RNAP, and cluster 1 was better than the other clusters. None of the 3 approaches to subsetting predicted time to PH. Subsetting based on Leroy classification and autoantibodies, but not clusters, predicted time to SF-36 PCS < 40, with dcSSc worse than lcSSc, and ATA worse than ACA.

Conclusion: Different approaches to subsetting provide different prognostic information. Subsetting based on clinical and serological profiles remains a challenge in SSc. In the future, subsetting based on molecular profiles may improve the predictive ability of SSc subsets.

	Mortality				FVC<70% predicted				ILD				PH				SF-36 PCS < 40
	N	N of events	Time to event (years)	P	N	N of events	Time to event (years)	P	N	N of events	Time to event (years)	P	N	N of events	Time to event (years)	P	N	N of events	Time to event (years)	P
Leroy subsets
dcSSc	384	66	13.0	ns	251	30	12.7	*	215	68	11.8	*	258	18	12.4	ns	145	69	12.0	*
lcSSc	350	50	13.9		228	15	13.4		229	34	13.1		233	19	13.3		126	36	12.8
Serology
ACA	250	31	14.2		176	14	14.0		185	26	13.6		162	10	13.3	ns	102	37	12.4
ATA	107	16	11.6		68	21	11.7	*	46	17	10.1	*	68	5	11.1		40	18	9.6	*
RNAP	127	29	11.9	*	80	9	11.0	*	74	27	10.6	*	87	8	11.5		37	17	12.3
Clusters#
Cluster 1	240	26	14.1		173	3	14.0		187	27	13.7		159	11	13.3	ns	103	37	12.6	ns
Cluster 2	237	44	12.1	*	161	19	11.8	*	153	44	11.3	*	174	15	11.7		91	36	11.9
Cluster 3	268	49	14.0	*	150	24	13.5	*	108	32	12.1	*	163	11	13.7		79	32	12.5

ns – not statistically significant; * p value < 0.05;ACA is the reference group; # Cluster 1 is the reference group

---

« Back to 2014 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/comparison-of-systemic-sclerosis-subsets-as-predictors-of-mortality-and-morbidity/