PJP incidence analysis in IIM cohort
14 PJP patients was found in IIM cohort (n=463) after one-year follow-up. All IIM patients was divided by PJP+ group and PJP- group to compare the risk factors of PJP infection. (table 1) We found that anti-MDA5+ is a risk factors for PJP infection in IIM (85.7% VS. 33.0%, P<0.0001), We calculated the incidence rate of PJP in IIM cohort is 3.0/100 person-year, while in MDA5+DM patients is 7.5/100 person-year and in MDA5-IIM patients is 0.7/100 person-year. (Figure 1 7.5% VS. 0.7%, P<0.00001) Besides that, factors like shorter median course (2 months VS. 6 months); interstitial lung (92.9% VS. 70.4%); complicated with diabetes (42.9% VS. 13.6%); higher prednisone exposure (50mg VS. 30mg), and laboratory examination differences such as higher erythrocyte sedimentation rate and ferritin, lower serum albumin, CD4+T lymphocytes counts, and lymphocytes (P < 0.05) may also contribute to PJP occurrence as well. In consideration that some of the factors mentioned above are characteristics of MDA5+DM, we further divided MDA5+DM patients into PJP+ group and PJP- group to confirm PJP infection factors. The Supplementary table S1(online) shows that PJP occurred in a median time of 2 months and with obvious decrease of CD4+ T cell counts and lymphocytes, which suggests that we may need to be on guard against the occurrence of PJP for MDA5+DM patients in the first 2 months of the disease or in the time when lymphocytopenia occur or CD4+ T cell counts decreases.
PJP+ cohort with rheumatic diseases
To evaluate the impact of anti-MDA5 (+) phenotype on mortality of PJP, we further analyzed all PJP+ patients (n=30) admitted into our medical center during the same period of the IIM cohort. The clinical features between MDA5+DM and other rheumatic disease were compared in Supplementary table S2(online), more details of 30 PJP+ patients in Supplementary table S3(online).
As shown in Figure 2A, The MDA5+DM (12, 39.60%) constituted the largest part of PJP+ patients, followed by Systemic Lupus Erythematosus (7, 22.70%), MDA5-IIM (2, 6.93%), ANCA associated vasculitis (AAV; 2, 6.93%), and Adult-Onset Still’s disease (AOSD; 2, 6.93%), primary Sjogren's Syndrome (pSS; 2, 6.93%), Undifferentiated connective tissue disease (UCTD; 2, 6.93%) and rheumatoid arthritis (RA; 1, 2.97%). The heatmap Figure 2B shows the rheumatic disease duration when PJP infection occurred. For IIM, AAV, AOSD patients, PJP infection most happened within 6 months of disease duration; while a little unexpectedly for SLE patients, PJP seemed to occur at any stage of disease in this cohort.
PJP mortality analysis in PJP cohort
The mortality of anti-MDA5-ab-positive was lethally higher than that of other rheumatic diseases (83.3% versus 38.9% P=0.016) as shown in Figure 2C and D. For 3-month-mortality risk factors, we identified age, CD4+ T cell counts and MDA5+DM with p values less than 0.10 in univariate analysis and then put the three factors in the cox regression model. It turned out that MDA5+DM and CD4+ T cell counts were identified as independent risk factors for death by multivariate analysis. Notably in these two, though as traditional PJP mortality factor [10], CD4+ T cell counts only had hazard ratio of 0.994 (95% CI 0.989-1.000) while MDA5+DM had HR of 3.254 (95% CI 1.209-8.756). (Table 2)
Besides baseline risk factors, prompt anti-PJP treatment is critical for patient survival. It has been long noticed that early diagnosis and early treatment can improve the prognosis of PJP patients [11][12]. We try to find the period-limitation for PJP-treatment by ROC curve analysis, but the cut-off valve of was not statistically significant (Supplementary figureS1A, P=0.0983, cut-off valve=7days). When we further stratified all the patients by anti-MDA5, a similar ROC curve as previous reports showed in patients with rheumatic diseases other than MDA5+DM. The time-to-PJP-treatment cut-off point of 6 days showed 85.7% sensitivity and 63.6% specificity, and with the Area Under Curve (AUC) 81.2% base on ROC curve. The time of six-day was the optimal cut-off point for timely treatment for PJP (Supplementary figureS1B, P=0.0297). However, in MDA5+DM patients, the ROC curve did not show at all (Figure S1C). We then analyzed patient survival to confirm the cut-off value. Consistent with previous reports, patient with other rheumatic disease tended to have better survival if they received anti-PJP treatment within 6 days after initial symptoms (Figure 3A, 1-year survival rate: 87.5% VS. 40%, P=0.057); while in MDA5+DM patients this phenomenon did not appear at all (Figure 3B, P=0.327). Timely-anti-PJP-treatment did not benefit the survival of MDA5+DM patients in our center.