Longitudinal Survival Outcomes in Allogeneic Stem Cell Transplantation: An Institutional Experience

Simple Summary Stem cell transplantation from a donor can be a potential cure for several blood cancers. We have maintained the characteristics and outcomes of patients receiving these transplants at our institution, the Ohio State University, since 1984. In this study, we looked at these outcomes, including how long patients lived after transplant, how long before their cancer came back, and if they developed other complications related to receiving a transplant, and evaluated how they changed over the years. We conclude that patients are now living longer with a lower chance of their cancer returning or of developing other complications, likely as a result of improvements in supportive care for transplant patients. Abstract Allogeneic hematopoietic stem cell transplantation (allo-SCT) is a potentially curative treatment for many hematological disorders, but is often complicated by relapse of the underlying disease, graft-versus-host disease (GVHD), and infectious complications. We conducted a retrospective analysis on patients undergoing allo-SCT from 1984 to 2018 to better understand how survival has changed longitudinally with therapeutic advancements made to mitigate these complications. Method: We analyzed data from 1943 consecutive patients who received allo-SCT. Patients were divided into groups (gps) based on the year (yr) of transplant. Primary endpoints were overall survival (OS), progression free survival (PFS), and GVHD-free relapse-free survival (GRFS). Secondary endpoints were the cumulative incidences of grade II–IV and grade III–IV acute GVHD (aGVHD), chronic GVHD (cGVHD), and non-relapse mortality (NRM). Results: Our study found statistically significant improvements in OS, PFS, and GRFS. Five-year PFS among the groups increased from 24% to 48% over the years. Five-year OS increased from 25% to 53%. Five-year GRFS significantly increased from 6% to 14%, but remained relatively unchanged from 2004 to 2018. Cumulative incidences of grade II–IV aGVHD increased since 2009 (p < 0.001). However, cumulative incidence of NRM decreased since 2004 (p < 0.001). Conclusions: Our data show improved OS, PFS, and GRFS post allo-SCT over decades. This may be attributed to advances in supportive care and treatments focused on mitigation of GVHD and relapse.


Introduction
In the past 50 years, over one million hematopoietic stem cell transplants have been performed, with approximately 42% percent being allogeneic hematopoietic stem cell transplants (allo-SCTs) [1]. While allo-HCT provides the potential for cure for patients with hematologic disease who may not otherwise have one, it is not without significant toxicity transfusion. Conditioning regimen and type were included as defined by the Center for International Blood and Marrow Transplantation (CIBMTR) research definitions for myeloablative (MA) and RIC [19]. Transplant-related complications like infections, veno-occlusive disease (VOD), hemorrhagic cystitis, and acute and chronic GVHD were collected. The dates of last follow up and relapse were also collected. This study was approved by the clinical scientific review committee and institutional review board at the Ohio State University.

Endpoints
The primary objective of this study was to assess how survival changed at our institution from 1984 to 2018. The primary endpoints of overall survival (OS), progression free survival (PFS), and GVHD-free relapse-free survival (GRFS) were used to compare outcomes across transplant years. The composite endpoint GRFS was assessed as the absence of grade III-IV acute GVHD (aGVHD), chronic GVHD (cGVHD) requiring system treatment, relapse, or death [20]. The secondary endpoints were the cumulative incidences of grade II-IV and grade III-IV acute GVHD (aGVHD), chronic GVHD (cGVHD), and non-relapse mortality (NRM). Classification of aGVHD was based on the Glucksberg grade, while cGVHD was classified using standard definitions [21,22].

Statistical Analysis
The Mann-Whitney U test for continuous variables and chi-squared or Fisher's exact test for categorical variables were used to compare patient-, disease-, and transplant-related characteristics among the seven groups. The Kaplan-Meier method was used to calculate the probabilities of OS, PFS, and GRFS and they were compared using the log-rank test.
Owing to the differences in length of follow-up, considering the median follow-up of 3.5 years in the most recent transplant group, all of the survival outcomes were censored at 3 years in the analysis. Gray's test, accounting for competing risks, was utilized to estimate cumulative incidence rates for relapse, NRM, and acute and chronic GVHD. The competing risks for GVHD were relapse and death; the competing risk for relapse was death; and the competing risk for NRM was disease-related deaths. A p-value of <0.05 was considered statistically significant. Analyses were performed using Stata 16 (College Station, TX, USA).

Patient and Disease Characteristics
Our analysis included 1943 consecutive patients who received an allo-SCT at the Ohio State University from 1984 to 2018 (Table 1). The median age at time of transplant was 50.0 years (range: 18-76), with 59.6% of the patients being male (n = 1158) and 94.3% being Caucasian (n = 1830). Acute myeloid leukemia (AML) accounted for the majority of transplants, at 36.3% of transplants, followed by non-Hodgkin lymphoma (14.2%), acute lymphoid leukemia (11.8%), chronic myeloid leukemia (10.1%), and myelodysplastic syndrome (10.0%). Across the time periods, the distribution of gender and race was similar across the groups. However, we did observe a significant increase in age across time with median age starting at 31.0 years in 1984-1988 to 57.0 years in 2014-2018 (p < 0.001). Complete comorbidity index data were available from 2004 based on the hematopoietic stem cell transplant comorbidity scoring system (HCT-CI) developed by Sorror et al [23].

Relapse
The cumulative incidence of relapse increased across the years at both the 5-year and 10-year marks. The rate of 5-year relapse was as follows: 22% (1984)(1985)(1986)(1987)(1988) After adjusting for the covariates as described previously, similar results were found with significantly increased relapse in the last three groups (Figure 3). Notably, matched-unrelated/mismatch-unrelated showed significantly decreased risk for relapse compared with match-related donors (p = 0.004 and p = 0.019, respectively). Reduced-intensity conditioning predicted significantly higher risk for relapse compared with myeloablative conditioning ((p < 0.001); Figure 3, Table 4). After adjusting for the covariates as described previously, similar results were found with significantly increased relapse in the last three groups (Figure 3). Notably, matched-unrelated/mismatch-unrelated showed significantly decreased risk for relapse compared with match-related donors (p = 0.004 and p = 0.019, respectively). Reduced-intensity conditioning predicted significantly higher risk for relapse compared with myeloablative conditioning ((p < 0.001); Figure 3, Table 4).

Survival Trends across Time Periods
The median follow-up time was 31.3, 26.7, 22.2, 17.2, 12.9, 8.1, and 3.5 years, respectively, across the years 1984-1988, 1989-1993, 1994-1998, 1999-2003, 2004-2008, 2009- worse OS outcomes (p = 0.049 and p = 0.006, respectively), while only F-M mismatch was a factor in PFS outcomes (p = 0.003). Myeloablative compared with reduced-intensity conditioning did not influence outcomes in OS and PFS, but marginally predicted worse GFRS outcomes (p = 0.052), likely owing to its role on GVHD.    Cox proportional hazard models were each established for PFS, OS, and GRFS. Overall, these models confirmed the findings from the Kaplan-Meier analysis for PFS, OS, and GRFS. However, GRFS modeling showed no significant increased risk among the different donor types (i.e., match-related, matched-unrelated, mismatch-related, and mismatchunrelated), while OS modeling indicated significantly increased risk with haploidentical and mismatched-unrelated donors and PFS modeling showed increased risk in mismatchunrelated donors only (p = 0.022). Donor gender mismatch (M-F and F-M) predicted worse OS outcomes (p = 0.049 and p = 0.006, respectively), while only F-M mismatch was a factor in PFS outcomes (p = 0.003). Myeloablative compared with reduced-intensity conditioning did not influence outcomes in OS and PFS, but marginally predicted worse GFRS outcomes (p = 0.052), likely owing to its role on GVHD.

Discussion
Overall, our data show improved OS, PFS, and GRFS post allo-SCT over decades. We observed an improvement in the rates of several key toxicities including hemorrhagic cystitis, veno-occlusive disease (VOD), and infectious complications. One factor that likely contributed to the decrease in these toxicities was the increased use of RIC regimens, particularly in older patients. The use of high-dose cyclophosphamide in conditioning regimens has also decreased over time, which likely accounts for the reductions in the incidence of hemorrhagic cystitis [24,25]. Several factors likely contributed to lower rates of VOD. For one, the increased use of RIC regimens likely led to lower rates of VOD, as MA regimens have been shown to have higher rates of VOD. The utilization of ursodiol for VOD prophylaxis starting in the early 2000s also likely contributed to lower rates of VOD, as this agent has been shown to decrease the incidence of hepatic aGVHD [26] and hepatic veno-occlusive disease [27,28]. Infectious complications, including bacteremia, viremia, and fungemia, also decreased over the years, which can likely be attributed to the development of more potent infectious disease prophylaxis. Our institution also implemented more sensitive viral testing in the early 2000s, which led to earlier detection and, consequently, prevention of more severe infections. With the increase in peripheral blood stem cells as the preferred source in recent years, faster ANC engraftment may also play a role in the decreased rates of infectious disease [29].
While our study showed a decrease in several key complications over time, it demonstrated an overall increase in the rate of grade II-IV aGVHD, but a relative stability of grade III-IV aGVHD. Several other studies also examined trends in aGVHD over time. In a single-center study, Gooley et  We observed a significant improvement in NRM over the years. Similar results have been seen in other studies [12,14,15]. As we perform more haploidentical and mismatch-unrelated allo-SCTs, it is important that we continue to optimize management of aGVHD, as rates of aGVHD have been shown to be higher with these sources. Options for steroid-refractory aGVHD have greatly expanded over the years, but there still is no definitive treatment option [30] and it remains an area of active research. One key medication that was recently approved for steroid refractory aGVHD is ruxolitinib, which has been shown to achieve overall response rates of up to 60%. However, additional agents are needed for patients who are either refractory to or are unable to tolerate ruxolitinib [31].
Relapse of primary disease continues to be the biggest challenge with allo-SCT. Our study showed increased risk of relapse of the years after adjusting for covariates. While Penack et al. and Mcdonald et al. showed a decrease in relapse, the improvements were minimal [15,16]. Additionally, previous studies on the role of unrelated donors compared with related donors on relapse have been inconclusive, though hypothetically, the role of the graft-versus-disease effect should be protective against relapse in unrelated donor allo-SCT [32,33]. Our study demonstrated, in comparison with match-related, a reduced risk of relapse in match-unrelated and mismatch-unrelated donor allo-SCTs. However, the general improvements in NRM, OS, and PFS can likely be attributed to improvement in supportive measures, including GVHD and infection prophylaxis, as well as reduced toxicity of conditioning regimens. Moreover, with better treatment, most of the patients recover and continue to live well after weaning off steroids and other immunosuppressive treatments. In concordance with prior studies, both OS and PFS for allo-SCT improved across the years [15,16].
Our study also conducted an analysis using the novel GRFS composite endpoint, useful for understanding the ideal recovery period with a balance between disease cure and ongoing morbidity [20]. The one-year GRFS in recent years (2014-2018) continues to be disappointing for allo-SCT, with approximately one in five patients reaching the one-year mark with ideal recovery. Upon further multivariable analysis, our study showed that donor type did not significantly predict GRFS. A poor response status predicted worse GRFS outcomes, while reduced-intensity conditioning predicted better outcomes.
Overall, in the context of several other studies, our study contributes to foundational knowledge to guide future work in allo-SCT. While aGVHD increased over the years, we saw a decrease in NRM and improvements in both OS and PFS. With GRFS being a composite endpoint balancing both aGVHD and relapse, focus should be on relapse mitigation and development of better treatment regimens for acute GVHD. We are aware of the limitations of our study. The retrospective design could introduce selection bias into our patient cohort. We were limited by the data variables available. Additionally, the increase in aGVHD II-IV could be because of the fact that there was an improvement in documentation in more recent years. Continued efforts to evaluate quality of life following allo-HCT will also be essential to tailor treatment to optimize patient-reported outcomes and minimize morbidity. Both treatment of relapse and potential maintenance therapies following transplant to mitigate relapse are also important areas of research and are the subject of several clinical trials. Studies are needed on these mechanisms and interventions for relapse to bring lasting clinical change following allo-SCTs.