The clinical characteristics of the patients included in this study are reported in table 1.
The cohort was composed of 98 patients classified as having SSc according to the ACR/EULAR criteria (4). They were aged between 21 and 84 years, with a large prevalence of females (ratio 10/1), and a comparable number of patients having lcSSc and dcSSc (48 and 50, respectively). The NVC pattern (5) was defined as early, active and late in 16, 42 and 40 patients, respectively. Twenty four out of 98 patients experienced one of more IDUs before their enrolment in the study. Anti-centromere and anti-topoisomerase-I antibodies were positive in 42 and 50 patients respectively.
During the follow-up 38 out of 98 patients (38.8%) developed one or more new IDUs. The T0-NEMO scores were significantly more elevated in the patients who developed IDUs with respect to those who did not [median 14.5 (95% CI 11.0-21.5), and 4.5 (95% CI 4.0-6.0), respectively, p<.0001] (Figure 1a). As expected, the T0-NEMO scores were significantly higher in patients with NVC active pattern [median 13.0 (95% CI 10.4-16.8)] with respect to those with early NVC pattern [median 5.5 (95% CI 1.6-12.4), p<0.02], and late NVC pattern [median 4.0 (95% CI 1.0-6.0), p< 0.0001]. No difference of the T0-NEMO score values was recorded between patients with early and late NVC patterns (p=0.23). Finally, the T0-NEMO scores were significantly more elevated in patients with dcSSc in comparison with patients with lcSSc [median 12.0 (95% CI 9.0-15.0) vs 5.0 (95% CI 4.0-6-0), p<0,003].
The appearance of new IDUs was associated with the history of previous ulcers, and with the presence of the diffuse cutaneous variant of the disorder (Chi square 15.6, p<0.0001 and 7.8, p<0.01, respectively). A similar association was found with the presence of anti-topoisomerasi I antibodies that are notoriously closely related with the dcSSc variant of the disorder. Conversely, no association was found between the appearance of IDUs and disease duration or use of more aggressive vasoactive therapies (monthly iloprost infusion, oral bosentan or sildenafil) in addition to basal CCBs. IDUs development during the follow up was observed in 4/16 (25.0%), 20/42 (47.6%), and 14/40 (35.0%) patients with early, active, and late NVC pattern, respectively. No statistically significant difference of the IDU prevalence between these NVC groups was recorded.
A logistic regression model in which all the three variables independently linked with subsequent appearance of IDUs were taken into account (T0-NEMO score, history of previous IDUs, and dcSSc variant) showed that only the history of previous ulcers and T0-NEMO gave a significant contribution to the model, whilst the third variable was discharged. (see table 2).
The ROC curve obtained by plotting sensitivity and 1-specificity of the different T0-NEMO score values in identifying patients who developed IDUs is represented in Figure 2a. The AUC of this ROC curve was 0.79 (95% CI 0.69-0.86, p<0.0001).
A NEMO score of 12 or more showed a sensitivity of 83.3% (95% CI 71.5-91.7), and a specificity of 63.2% (46.0-78.2), with a PPV and NPV of 58.9% (95% CI 44.7-72.2), and 85.6% (71.8-94.4), respectively. Furthermore, a NEMO score of 16 or more was highly predictive of future development of IDUs showing, in this respect, a sensitivity of 95.0% (95% CI 86.1-99.0), and a NPV of 93.4% (95% CI 77.5-99.2).
The Kaplan-Meier curve analysis confirmed that the development of subsequent IDUs was significantly more frequent in patients having a T0-NEMO score of 12 or more with respect to those with a lower baseline NEMO score. The difference between the two curves became significant after 6 months from T0 (Figure 3a).
When the same statistical analysis was limited to the 74 patients who had no previous occurrence of IDUs, the results were very similar. Twenty-eight of them (37.8%) developed at least one IDU during the two-year follow-up. The T0-NEMO scores had a median value of 14.0 (95% CI 10.8-22.9) in the patients who developed IDUs, and of 4.5 (95% CI 3.9-6.3) in those who did not (p<0.0001] (Figure 1b). The AUC of the ROC curve (Figure 2b) was slightly higher [0.86 (95% CI 0.76-0.93), p<0.0001] than that found taking into consideration the totality of patients. This difference, however, was not significant (z value 1.01, p=0.3). In this more restricted cohort a NEMO score of 12 or more had a sensitivity of 80.4% (95%CI 66.1-90.6), and a specificity of 78.6% (95% CI 59.0-91.7), while a NEMO score of 16 or more maintained a strongly predictive value of future development of IDUs showing a sensitivity 93.5% (95%CI 82.1-98.6), and NPV of 93.1% (95% CI 76.2-99.3).
A Kaplan-Meyer curve was also built and analyzed in patients who were naïve for previous IDUs. Even in this case, the appearance of IDUs was more frequent in patients having a T0-NEMO score ≥12, and this difference was comparably significant at 6 and 24 months (Figure 3b).
The statistical analysis was also performed in patients with early and active NVC pattern, thus excluding those patients with late NVC pattern who had more pronounced fibrotic changes.
In this selected cohort the T0-NEMO scores had a median value of 24.0 (95% CI 15.0-34.5) in patients who developed IDUs and of 5.5 (95%CI 4.0-10.3) in those who did not (p<0.0001) (Figure 1c). The AUC of the ROC curve (Figure 2c) was again slightly higher [0.89 (95% CI 0.78-0.96), but not significantly different from that obtained in the whole group of patients (z value 1.45, p=0.14). A NEMO score of 16 or more was again very strongly associated with future development of IDUs, showing a sensitivity of 94.1 (95%CI 82.4-97.4) and a NPV of 94.1 (95%CI 76.4-99.6).