Autologous hematopoietic stem cell transplantation improves long-term survival—data from a national registry

Patients met the 2013 American College of Rheumatology (ACR)/European League against Rheumatism collaborative initiative criteria for systemic sclerosis [15]. The SSc subtype was based on the classification criteria established by LeRoy et al. [16, 17]. Follow-up visits were performed at least once per year. In the present analysis, we focused on male patients with dc-SSc and positive Scl70 antibodies as a control group.

The modified Rodnan skin sclerosis (mRSS) score was used to quantify skin sclerosis. Repetitive determination of the mRSS score was preferentially performed by the same trained investigator to minimize variations.

Heart involvement was defined as the presence of elevated troponin T, pericardial effusion, pathologic changes in cardiac MRI, ventricular extrasystoles > Lown 3, or the presence of endomyocardial fibrosis in a biopsy specimen.

Patients received HSCT between 1997 and 2018. Cyclophosphamide 2×2g/m² was used for mobilization chemotherapy until 2012 and 2×1g/m² after 2012 with equally sufficient hematopoietic stem cell (HSC) harvest gains. All patients received G-CSF from day +4 until the end of the HSC harvest. Some transplants were manipulated by enhancing HSC using CD34-loaded columns and/or anti-CD3 and anti-CD19 monoclonal antibodies to minimize residual autoreactive T- and B-cells in the transplant. In any case, the goal was to receive 2.5×10⁶ CD34+ HSC per kg body weight at the end of the harvesting procedure. Conditioning chemotherapy was performed by using cyclophosphamide 200 mg/kg, melphalan 140 mg/m², or a combination of cyclophosphamide 100 mg/kg and thiotepa 10 mg/kg in patients with cardiac involvement. After the hematologic reconstitution, a prophylaxis with cotrimoxazole, nystatin, and aciclovir was continued for 3 to 6 months to prevent pneumocystis, candida, and herpes virus reactivation. We did not recognize any significant differences between patients and outcomes and between the several protocols.

The DNSS data registry was analyzed for patients who were treated with HSCT. Two control groups of patients with dc-SSc were used as references. Control group A comprised patients with dc-SSc and moderate risk for mortality. Control group B comprised male patients with dc-SSc who were positive for anti-Scl70 and had a high risk for mortality. To compare demographic and serological parameters, we used data of the initial visit. To compare organ manifestations and HSCT treatment, we used follow-up data of all visits.

The overall survival was assessed using Kaplan-Meier analyses and compared between the HSCT group and control groups using log-rank tests. Results are presented as Kaplan-Meier curves and the corresponding 95% confidence intervals were compared 5, 10, and 15 years after SSc onset which is the time of the first non-Raynaud manifestation.

To investigate mortality risk factors for HSCT-treated patients, we compared patients younger than age 50 years with older patients and SSc disease duration of less than 3 years between the first non-Raynaud symptom and longer SSc disease duration.

For group comparisons, we used Pearson’s chi-square test (qualitative data) and the Kruskal-Wallis test (quantitative data). All reported p values are 2-sided and p values of < 0.05 were considered statistically significant. As the analyses were regarded as explorative, we did not adjust for multiple testing. Calculations and figures were carried out using SPSS (23.0.0.3 64-Bit, IBM Corp., Armonk, NY, USA) and Excel (Microsoft Corp., Redmond, WA, USA), as well as R (version 3.4.0, R Foundation for Statistical Computing, Vienna, Austria).

Between 2003 and 2019, more than 5000 patients with systemic sclerosis (SSc) were recruited to the German Network for Systemic Scleroderma (DNSS). Within this cohort, 80 patients (1.6%) were treated with high-dose chemotherapy followed by autologous hematopoietic stem cell transplantation (HSCT) between 1997 and 2018. Indications for HSCT were an mRSS > 10 (68%), presence of ILD (66%), heart involvement (24%), or combinations of poor risk factors. Nine patients received HSCT before the year 2003 and before they entered the DNSS registry. The mean age at SSc diagnosis, HSCT, and DNSS baseline were 38.3 years, 43.0 years, and 44.3 years, respectively (Table 1). The mean time period (± SD) between the age at SSc diagnosis and the age at HSCT was 4.1 ± 4.8 years. Sixty-nine patients (86.3%) in the HSCT group had a diffuse cutaneous (dc-SSc) form of SSc. Therefore, dc-SSc patients who did not receive HSCT were considered as control group A. Within the DNSS registry, 1513 patients with dc-SSC were identified as control group A (Table 1). However, further analyses indicated statistically significant differences between the HSCT and control group A. HSCT patients were younger at dc-SSc diagnosis and younger at DNSS entry at baseline (Table 1). In the HSCT group, more patients were male (43.8% versus 26.8%), had anti-Scl70 antibodies (71.3% versus 53.7%), and had a lower body mass index (BMI 22.8 kg/m² versus 24.3 kg/m²) compared to control group A (Table 1). These findings confirm that SSc patients had a poor prognosis before they entered the HSCT program. The epidemiologic findings in Table 1 also suggest that all patients with dc-SSc (control group A) might be not an adequate control group for HSCT.

Therefore, we decided to include a second control group comprising patients with dc-SSc and Scl70 antibodies and male gender (control group B). Among control group A, 240 patients were identified for the more restrictively defined control group B (Table 1). Although control group B was defined as a more homogenous control group with high-risk features, significant differences regarding the age at SSc diagnosis, age at DNSS entry, ANA and Scl70 positivity, and BMI persisted when compared with the HSCT group (Table 1).

Quantification of the skin sclerosis with the modified Rodnan skin score (mRSS) was 17.7 on average in the HSCT group, 15.6 in the control group A, and 17.2 in the high-risk control group B (Table 2). We found a significantly lower number of patients with severe skin sclerosis (mRSS > 20) in control group A (28.2%) compared to the HSCT group (41.0%) and control group B (32.1%).

ILD was observed in 52.5% of the HSCT group, compared to 43.5% in control group A and 47.1% in control group B (Table 2) confirming that patients in the HSCT group were negatively selected for rapidly progressive disease. Since PAH has a well-known high risk for mortality, patients with unstable or moderate to severe PAH should not be considered for HSCT. Eleven patients with mild and stable PH associated with ILD (13.8%) were identified in the HSCT group compared with 11.3% in control group A and 11.7% in control group B (Table 2). The single-breath diffusion lung capacity for carbon monoxide (DLCO-SB) was more impaired in the HSCT group (53.9%) compared with control group A (63.2%, Table 2).

Heart involvement was documented in 23.4% in the HSCT group, 16.1% in control group A, and 21.7% in control group B, which confirms that control group B is a better comparator for patients in the HSCT group.

Between 1997 and 2012, 37 patients received mobilization chemotherapy with cyclophosphamide 4 g/m² which was applied on two consecutive days (Table 2). During 2012, the intensity of the mobilization chemotherapy was reduced to cyclophosphamide 2 g/m² with sufficient yield of the stem cell harvest. Three patients received plerixafor to increase the yield of harvested stem cells. CD34 columns were used in 76 grafts to concentrate hematopoietic stem cells and to remove potentially autoreactive lymphocytes from the graft.

Various protocols were used for conditioning chemotherapy according to the local standards (Table 3). The protocol of the ASTIS trial was considered as the standard protocol because the majority of transplant centers also participated in ASTIS [10].

The age of the participants is usually restricted in clinical trials. We analyzed patients, who were younger than 50 years (mean 38.5 years, 76%) or older than 50 years (mean 57.3 years, 24%) when HSCT was performed (Table 4 part A). Older patients had a higher risk for ILD and PAH (33% versus 9%, p=0.025). The overall survival decreased from 95 to 83% when HSCT was performed before or after the age of 50 years (Table 4 part A). This suggested a higher mortality in patients with the diagnosis of SSc after age 50 years and rapidly progressive disease. Only 6 deaths (7.5%) were observed in this cohort of 80 patients with HSCT during the follow-up between mean 4.3 and 6.9 years (Table 4 part A).

Another restriction in clinical trials is the definition of early disease, which is usually defined as 3 years from the first non-Raynaud symptom to study baseline. In our HSCT cohort, 56% of patients were treated earlier than 3 years (mean 1.6 years) with HSCT compared to 44% of patients later than 3 years (mean 7.4 years, Table 4 part B). Since the age at HSCT was comparable in both groups (43.2 versus 42.9 years), this showed that a rapid progression of skin sclerosis usually occurs in older patients who are rapidly referred to an HSCT center. Five patients died in the HSCT group earlier than 3 years and 1 when HSCT was performed later than 3 years of SSc diagnosis. These findings confirm the strong selection of referral and HSCT centers for patients with multiple risk factors for rapidly progressive disease and considerable risk of TRM.

The DNSS registry comprises long-term follow-up of SSc patients for up to 15 years (Fig. 1). A cumulative survival of 92% of HSCT patients was observed after 15 years of follow-up. Patients who did not receive HSCT seemed to live longer within the first 5 to 10 years after SSc diagnosis (Fig. 1A). The survival seemed to remain stable in HSCT and seemed to decline continuously in control group A and control group B during the 15 years of follow-up. However, the absolute numbers were getting small at the end of the long-term follow-up as indicated (Fig. 1). A more rapid decline of cumulative survival was observed in control group B (Fig. 1B). The 5-, 10-, and 15-year survival rates were 96%, 92%, and 86% in control group A (Fig. 1A) and 93%, 81%, and 71% in control group B which was significant after 15 years (p=0.041, Fig. 1B). The decline of the survival within the first 5 years of observation is associated with rapidly progressive SSc but not with HSCT or other treatments (Fig. 1B). The decline of patient numbers after 5-, 10-, and 15 years seemed to be only marginally attributed with a loss to follow-up. In the combined control group A and HSCT cohort, only 15 (3.0%), 25 (3.0%), and 41 (3.7%) patients were actually registered as lost to follow-up.

In the HSCT group, one of 80 patients died within 100 days after HSCT resulting in a TRM of 1.3% in this cohort. Additional 5 patients died later than 100 days post-HSCT due to progressive SSc. If we consider these patients as non-responders, we find a non-response rate of 6.3% after HSCT. This has to be compared with 125 patients (10.4%) who died in control group A and 31 patients (12.9%) who died in control group B.

Since the absolute numbers declined during follow-up, we focused on the first 5 years of follow-up in the DNSS registry. Because of an average delay of about 5 years between diagnosis and DNSS entry, the follow-up of years 0 to 5 in Table 3 refers to the years 5 to 10 after SSc diagnosis. We observed a significant decline of mRSS of 17.6 (± 11.5) to 11.0 (± 8.5) during the following 5 years. Patients in the high-risk control group B had a higher mRSS and lower DLCO-SB at baseline without a significant improvement during follow-up for 5 years (Table 5).

If we consider a minimal detectable mRSS improvement of −2 points, we find at least minimal improvement of skin sclerosis in 48.0% of patients 1 year after HSCT. In control group B, we detected a minimal skin improvement in 28.4% of patients not earlier than 5 years after baseline (data not shown). These data suggest that 52.0% of patients do not show improvement of skin sclerosis or were transplanted for other reasons, e.g., heart or lung disease or an accumulation of several poor risk factors. If the response to HSCT was defined as a detectable improvement of skin sclerosis, and not only reduction of cumulative risk, we have to accept that about 30% of patients will not achieve an improvement of skin sclerosis (data not shown). Results from clinical trials estimated a minimally important difference in mRSS score in the range between 3.2 and 5.3 [18]. If we consider an mRSS response of −4 points as clinically relevant improvement, the mRSS response to HSCT would be lower than 40%.