Adult umbilical cord blood transplantation (CBT) has emerged as an important option for patients lacking matched related (MRD) and matched unrelated donors (MUD). We compared chronic GVHD (cGVHD) incidence, immunosuppression burden and late infections and hospitalizations in consecutive patients undergoing CBT (n=51) versus peripheral blood MUD transplant (n=57) at our center between June 2009 and April 2014. At 3 years post transplantation, the cumulative incidence (CI) of moderate to severe cGVHD was 44% following MUD versus 8% following CBT (P=0.0006) and CI of any cGVHD was 68% following MUD versus 32% following CBT (P=0.0017). Median time to being off immunosuppression among CB patients was 268 days versus not reached among MUD patients (P<0.0001). Late infections and late hospitalized days were reduced in CB patients (P=0.1 and <0.001, respectively). Three-year CI of transplant-related mortality (TRM) and relapse as well as 3-year overall survival (OS) were similar following CB and MUD transplantation. We demonstrate a significantly lower incidence of cGVHD, immunosuppression burden and late complication rate following UCB versus peripheral blood MUD transplant without decreased OS, increased relapse or early TRM. Combined with the rapid availability of UCB, these findings have led our center to move primarily to UCB over peripheral blood MUD when a MRD is not available.
Umbilical cord blood (CB) has emerged as an important donor source for patients lacking matched related (MRD) and matched unrelated donors (MUDs). Numerous recent studies comparing transplantation outcomes using each donor source have demonstrated similar overall survival (OS) in both adult and pediatric populations.1, 2, 3, 4, 5, 6 However, owing to ongoing concerns about graft failure, delayed engraftment, delayed immune reconstitution and costs following CB transplantation (CBT), most adult centers currently prioritize MUD over CB.
Among long-term relapse-free survivors of transplantation, chronic GVHD (cGVHD) and its sequelae account for the majority of post-transplant morbidity and mortality. We compared outcomes of consecutive adult peripheral blood MUD and CBT at our center with a focus on cGVHD as well as late infections and hospitalizations. Whereas OS, relapse, and both early and late transplant-related mortality (TRM) were comparable between each group, the incidence of cGVHD, time to immunosuppression withdrawal and incidence of late post-transplantation infections and hospitalizations were lower following CBT versus peripheral blood MUD transplantation.
Patients and methods
We compared outcomes in consecutive patients undergoing first CB (n=51) versus first peripheral blood MUD transplantation (n=57) at our center between June 2009 and April 2014. No bone marrow transplants were performed for hematologic malignancies during this period. All patients signed IRB-approved informed consent authorizing data collection for this study. Patients were treated on multiple protocols including research and standard of care regimens. Acute GVHD (aGVHD) was graded according to established criteria.7 cGVHD and late aGVHD was graded according to the 2005 National Institutes of Health (NIH) consensus criteria.8 Patient and transplant details are summarized in Table 1. Among CB patients, median age at transplant was 58 (range 22–73) versus 54 (range 18–72) among MUDs. Eighteen CB patients underwent myeloablative conditioning and 33 patients underwent non-myeloablative conditioning for ALL (n=7), AML (n=27), myelodysplastic syndrome (n=5), CLL (n=3), non Hodgkins lymphoma (n=7) and CML (n=2). GVHD prophylaxis was cyclosporine and mycophenolate mofetil. All CB patients received double cord transplants. Median cryopreserved total nucleated cell dose per unit was 2.3 × 107/kg (1.5–5.1). Matching was 4/6, 4/6 n=9, 4/6, 5/6 n=11, 5/6,5/6 n=20, 6/6,5/6 n=5, 6/6,6/6 n=4, 6/6/, 4/6 n=2. Fifty-four MUD patients were 10/10 matched and three patients were 8/8 matched (DQ mismatched). Eighteen patients underwent myeloablative conditioning (tacrolimus/methotrexate GVHD prophylaxis) and 39 patients underwent non-myeloablative conditioning (tacrolimus/mycophenolate mofetil GVHD prophylaxis) for ALL (n=8), AML (n=13), non Hodgkins lymphoma (n=15), myelodysplastic syndrome (n=10), CLL (n=5) and other (n=6). For all patients, standard immunosuppression taper in the absence of GVHD or relapse was scheduled to be completed approximately 6 months following transplant. Through November 2013, our donor selection algorithm prioritized MUDs over CB, and all CB transplants were performed on patients lacking available MUDs. On the basis of preliminary analyses of the data presented in this manuscript, our donor selection algorithm shifted to prioritizing CB over MUD transplant in November 2013. Between November 2013 and April 2014, 13 CB and 2 MUD transplants were performed. As a result, median follow-up is shorter in the CB population. However, all results are similar to analyses performed on patients transplanted between June 2009 and November 2013.
Patient characteristics and follow-up times are summarized using standard measures. Cumulative incidence (CI) estimates of aGVHD and cGVHD, time to immunosuppression withdrawal, relapse and TRM were utilized, with relapse, TRM, immunosuppression taper for persistent disease and graft failure (defined as sustained trilineage cytopenias without other explanation or loss of detectable donor chimerism) included as competing risk events as appropriate.9 For aGVHD, cGVHD and time to immunosuppression withdrawal, relapse, TRM and graft failure were considered competing risks. For relapse, TRM was considered a competing risk, and for TRM, relapse was considered a competing risk. CI curves were compared using Gray’s test. Kaplan–Meier estimates were used to evaluate OS. Censoring for all time-to-event outcomes occurred at the date of last contact. Kaplan–Meier curves were compared using log-rank test.
For assessments of infectious episodes and hospital days/days of follow-up, a Wald test statistics was constructed to compare the two groups. Ninety-five percent confidence intervals (CIs) were calculated for the relative risks.
Landmark analysis beginning day 90 post transplantation was used to assess CI of cGVHD as well as late infections and hospitalizations. Landmark analysis beginning day 180 post transplantation was used to assess CI of time to immunosuppression withdrawal. Statistical analysis was performed using SAS statistical software.
Twenty CB patients experienced grade 2–4 aGVHD prior to day 100 (two grade 3–4) and six (all of whom experienced aGVHD prior to day 100) experienced recrudescent aGVHD after day 100 (including one patient who escalated from grade 3 to grade 4). Twenty-two MUDs experienced grade 2–4 aGVHD prior to day 100 (7 grade 3–4) and 12 (4 of whom did not experience aGVHD prior to day 100) experienced late aGVHD (including one patient who escalated from grade 2 to grade 3). CI of grade 2–4 aGVHD was 39% for CB patients and 47% for MUD patients (P=0.47) (Figure 1).
At 3 years post transplant, CI of moderate-to-severe cGVHD was 44% following MUD versus 8% following CBT (P=0.0006) and CI of any cGVHD was 68% following MUD versus 32% following CBT (P=0.0017) (Figure 2). The CI of immunosuppression withdrawal was significantly more rapid following CB than MUD transplant (P<0.0001); median time to being off immunosuppression among CB patients was 268 days versus not reached among MUD patients (Figure 3). Median follow-up among all CB patients was 411 days (range 18–1853) and median time to onset of cGVHD was 210 days (range 108–490). Onset was de novo n=3, quiescent n=8 and progressive n=3. Median follow-up among all MUD patients was days 488 days (range 24–1955) and median time to onset of cGVHD was 225 days (range 83–803). Onset was de novo n=7, quiescent n=12 and progressive n=11.
Among CB patients, one patient remained on immunosuppression 384 days post transplant. One patient who died of cGVHD-related complications remained on immunosuppression at the time of death. Four patients restarted immunosuppression owing to recrudescent symptoms, but all four patients subsequently tapered again and remain off immunosuppression. Among MUD patients, the five patients who died of cGVHD-related complications and the two patients who died of relapse after developing cGVHD remained on immunosuppression at the time of death. Twenty-two of 31 surviving patients who did not experience competing risks prior to onset of cGVHD remain on immunosuppression without ever stopping. Of nine patients who stopped immunosuppression, eight subsequently restarted for recrudescent GVHD symptoms and five remain on immunosuppressive therapy.
Late infections and hospitalizations
We performed a descriptive analysis comparing late infection and hospitalization rates among CB and MUD patients who underwent successful transplants without early competing risks and without late relapse. Analysis began 90 days post transplantation; patients who died or relapsed (n=8 CB, 10 MUD), experienced graft failure (n=3 CB, 5 MUD) or underwent rapid immunosuppression for persistent disease (n=1 CB, 1 MUD) prior to day 90, as well as any patients experiencing the competing risk of relapse (n=8 CB, 7 MUD) after day 90 post transplantation were excluded. Infectious episodes were defined as bacterial, fungal or viral infections in which patients were symptomatic or required treatment (CMV reactivation) and pathogens were identified by culture, PCR or histological exam. Among 31 CB patients (median follow-up 551 days (range 40–1763)) and 34 MUD patients (median follow-up 920 days (range 37–1865)), comprehensive follow-up data were not available for 1 MUD patient. Six cord patients died (cGVHD=1, aGVHD n=1, infection n=2, pulmonary failure n=2) and nine MUD patients died (cGVHD n=5, aGVHD n=1, suicide n=1, infection n=1, pulmonary failure n=1) during follow-up.
Of the CB patients, 12 did not experience infectious episodes after 3 months post transplantation and 19 experienced infections (median 2 infectious episodes, range 1–11). Of the 33 assessable MUD patients, 4 did not experience infectious episodes after 3 months post transplantation and 29 experienced infections (median 3 infectious episodes, range 1–9). The relative risk of late infection episode on a per infection level (total number of infections, including multiple infections in a single individual) for CB patients was comparable with MUD patients (P=0.37, relative risk 0.95 (95% CI 0.67–0.1.3)). Excluding patients who died during the follow-up period, the relative risk of late infection episode on a per infection level for CB patients trended toward lower compared with MUD patients (P=0.10), relative risk 0.77 (95% CI 0.51–1.16). (Table 2)
Among CB patients, 16 were not rehospitalized after 3 months post transplantation and 15 were readmitted (median 2 readmissions, range 1–10). For 19 405 total follow-up days, patients spent 535 days hospitalized (0.028 hospital days/follow-up days). Among MUD patients, 9 were not rehospitalized after 3 months post transplantation and 25 were readmitted (median 1 readmissions, range 1–7). For 29 418 total follow-up days, patients spent 825 days hospitalized (0.028 hospital days/follow-up days). The relative risk of late hospitalized days following CBT was comparable with that following MUD transplantation (P=0.4, relative risk 0.98 (95% CI 0.88–1.1). Excluding patients who died during the follow-up period, the relative risk of late hospitalized days following CBT was lower than following MUD transplantation (P<0.001), relative risk 0.74 (95% CI 0.66–0.82). (Table 2)
Given the variable follow-up time among patients included in the analyses of infections and hospitalizations above, we assessed infections and hospitalizations for all surviving patients between days 90 and 309 post transplantation (309 days being the shortest follow-up). Results were unchanged with this analysis.
Relapse, TRM and OS
Among the 51 CBT patients, there were 21 deaths due to relapse (n=11) or TRM (n=10). Among patients experiencing TRM, causes of death included cGVHD (n=1), aGVHD (n=1) conditioning-related toxicity (n=1), infection (n=3), late pulmonary failure (n=2) and graft failure with subsequent relapse (n=2). Twelve patients experienced relapse. Among the 57 MUD patients, there were 25 deaths due to relapse (n=10) or TRM (n=15). Among patients experiencing TRM, causes of death included cGVHD (n=5), aGVHD (n=2), conditioning-related toxicity (n=2), infection (n=1), late pulmonary failure (n=1), graft failure with subsequent relapse (n=2) or death due to GVHD following subsequent transplant (n=1), and suicide (n=1). Sixteen patients experienced relapse. Three-year CI of TRM was 25% for CB and 24% for MUD patients (P=0.73) and 3-year CI of relapse was 22% for CB and 24% for MUD patients (P=0.86). Three-year OS was 54% for CB and 52% for MUD patients (P=0.68). (Figure 1)
cGVHD is the most important cause of long-term morbidity and mortality among non-relapsed survivors of HSCT. Our data confirm previous reports suggesting a low incidence of cGVHD following CBT as compared with peripheral blood URD transplant.1, 2, 10 In addition, we demonstrate significantly more rapid removal of immunosuppression and sustained immunosuppression discontinuation following CBT as compared with peripheral blood MUD transplantation as well as decreased infections and hospital days after more than 90 days post transplantation. These data have important implications for long-term survivors of transplantation; if OS and other complications are comparable between the two groups, being alive without cGVHD or on immunosuppression may lead to improved quality of life and reduced costs of care.
The 3-year CI by landmark analysis of any as well as moderate/severe cGVHD was significantly lower following CBT as compared with MUD transplantation (32 versus 68%, P=0.0017 and 8 versus 44%, P=0.0006, respectively). Owing to censorship of patients experiencing competing risks, CI of cGVHD following transplant has some limitations, though it is the standard method for calculating incidence of cGVHD. By also assessing cGVHD incidence in patients not experiencing competing risks of relapse, TRM or graft failure prior to the onset of cGVHD, we are able to predict the likelihood of being alive with cGVHD following an otherwise successful transplant. When counseling patients regarding risks of cGVHD, this information may be more valuable then describing the landmark cumulative incidence analysis risk of cGVHD. In our study, though the CI of cGVHD among MUDs was 68% percent, 94% of MUD patients who were alive without competing risk experienced cGVHD (63% moderate to severe) in contrast to a 32% CI of cGVHD among CB patients and only a 52% chance of cGVHD among CB patients (11% moderate to severe) who were alive without competing risk.
In addition, our data demonstrate time to immunosuppression discontinuation, as well as sustained immunosuppression discontinuation, markedly favor CBT. Whereas all of the CB patients were off immunosuppression by 500 days post transplantation, 84% of surviving MUD patients remained on immunosuppression 2 years post transplantation. Though the median dose was low in these MUD patients (10 mg of prednisone and low dose calcineurin inhibitors), side effects, costs and inconvenience of the medications remain important issues independent of symptoms related to cGVHD. Time to immunosuppression discontinuation data can also be deceptive, as many studies do not indicate whether immunosuppression needs to be restarted following tapering to off. Notably, four CBT patients who tapered immunosuppression restarted for recrudescent symptoms, but all four rapidly re-tapered and remain off immunosuppression. In contrast, 8 of 10 MUD patients who tapered off immunosuppression subsequently restarted, and all remain on immunosuppression.
We also assessed late infections and hospitalizations following transplantation and found fewer events in the CBT group. Others have reported excellent long-term outcomes among CBT patients not experiencing early TRM,11, 12 and our data are consistent with these findings. As expected, the highest burden of late hospitalization and infection fell on patients with significant cGVHD. Among long-term survivors, late hospitalization rates were lower following CBT and infection rates trended toward fewer (P=0.1) following CBT.
Adult CBT is historically associated with several concerns in the early post-transplant period including higher rates of graft failure, delayed engraftment and higher risk for infection early post transplantation. Optimization of conditioning regimens including development of middle intensity regimens,13 use of double unit transplants to decrease graft failure,14, 15 ex vivo expansion and other strategies to decrease time to engraftment,16, 17 and accumulation of an increasing inventory of CB units allowing for selection of larger and better matched units have all contributed to decreases in graft failure and lower early post-transplantation infection risk. In our series, early TRM was comparable between CBT and MUD donors.
aGVHD, the other significant source of early post-transplant morbidity and mortality, remains an issue following CBT as well as MUD transplantation. In our series, the incidence was comparable following transplant with each donor source. This finding is consistent with other reports.1, 2 However, although the incidence of aGVHD was comparable between donor sources, others have suggested higher sensitivity of CB patients to aGVHD treatment.18 Our data are consistent with this observation and is reflected in the successful rates of rapid sustained immunosuppression withdrawal. Others have also noted late aGVHD to be common following CBT.18 Although we observed more late aGVHD than classic cGVHD in our cord patients, the issue was not significant, as again reflected in the successful rates of rapid sustained immunosuppression withdrawal.
Our study is small, retrospective and involves heterogeneous populations. Importantly, our study compared CBT with peripheral blood MUD transplantation; incidence of cGVHD following bone marrow transplantation or T-cell-depleted transplantation is known to be lower. Several different conditioning regimens and GVHD prophylaxis strategies were used and multiple diseases at different stages were included. However, all GVHD prophylaxis regimens have been well described in conjunction with their corresponding conditioning regimen and donor source. The CBT and MUD populations were well matched with respect to age, disease, disease status, comorbidity index and proportion of myeloablative and non-myeloablative transplants.
Others have reported excellent long-term outcomes for CB patients who survive the acute transplant period. Our data are consistent with these observations, and, importantly, we did not observe increased early TRM following CBT. The decreased cGVHD and late complications we have observed following CBT combined with the rapid availability of CB has led our center to move primarily to CB when a MRD is not available, particularly when large and well-matched CB units are available. Ongoing analysis will examine whether moving directly to CBT for patients with common haplotypes with MUDs available will result in larger, better matched CB units for patients and whether this will result in improved outcomes. Additional assessments of late and total costs of care (including potential cost reductions resulting from more rapid transplantation with less consolidation) following CBT as well as formal quality of life assessments are ongoing.
About this article
Thiotepa-Based Intensified Reduced-Intensity Conditioning Adult Double-Unit Cord Blood Hematopoietic Stem Cell Transplantation Results in Decreased Relapse Rate and Improved Survival Compared with Transplantation Following Standard Reduced-Intensity Conditioning: A Retrospective Cohort Comparison
Biology of Blood and Marrow Transplantation (2018)