Volume 14, Issue 12 , Pages 1380-1384, December 2008
Treatment Change as a Predictor of Outcome among Patients with Classic Chronic Graft-versus-Host Disease
Article Outline
Abstract
We analyzed outcomes for 668 patients who had systemic treatment for chronic graft-versus-host disease (cGVHD) to assess the utility of early treatment change for exacerbation of cGVHD as a surrogate for survival endpoints in clinical trials. Fifty-six percent of patients had treatment change within 2 years after diagnosis of cGVHD. The median onset of treatment change was 4.4 months (range: 0.3-50 months). The cumulative incidence of nonrelapse mortality (NRM) at 2 years was 16%, and overall survival (OS) at 2 years was 74%. In time-dependent Cox models, treatment change was associated with an increase in risk of NRM (hazard ratio, 2.53; 95% confidence interval, 1.7-3.7; P < .0001). The hazard ratio was attenuated by 6% per month of delay in treatment change. Our results confirm that exacerbation of cGVHD is associated with an increased risk of NRM and with decreased OS, but the strength of this association is not large enough to allow the use of early exacerbation as a surrogate for survival endpoints in clinical trials. Other measures of clinical benefit, such as response, will need to be developed as endpoints in phase II trials for patients with cGVHD.
Key Words: Chronic graft-versus-host disease, GVHD
Introduction
Chronic graft-versus-host disease (cGVHD) is the major determinant of both late nonrelapse morbidity and mortality (NRM) after allogeneic hematopoietic cell transplantation (HCT) 1, 2, 3. In addition to increased risk of NRM, patients with cGVHD have decreased performance, impaired quality of life, and delayed immune reconstitution 4, 5, 6. In most cases, resolution of cGVHD requires prolonged systemic immunosuppressive treatment [7].
Clinical trials of treatment for cGVHD have been hampered by the lack of easily documented early indicators of clinical benefit. Demonstration of prolonged survival or a shorter time to resolution of cGVHD may be feasible for pivotal phase III studies, but ascertainment of these endpoints occurs too late to be useful for phase II studies 8, 9. A National Institute of Health (NIH) consensus development project recommended an approach that could be used to measure response in clinical trials for treatment of cGVHD, but these recommendations have not yet been validated [10]. Implementation of these recommendations has been difficult, because they require a much larger burden of documentation than would ordinarily be required for clinical practice. In addition, certain features of cGVHD have important qualitative characteristics such as sclerosis and intensity of erythema that are not readily amenable to quantitative measurement.
Identification of more accessible validated endpoints indicating clinical benefit could facilitate the conduct of clinical trials in cGVHD. Kim et al. [11] recently proposed that an early flare or exacerbation of cGVHD could be used as an endpoint in clinical trials. Early flare was strongly associated with decreased GVHD-specific survival, that is, increased NRM. GVHD-specific survival at 2 years was 96% among 27 patients who had no flare, 65% among 34 patients with a first flare more than 78 days after diagnosis of cGVHD, and 33% among 35 patients who had a first flare within 78 days after diagnosis [11]. Prompted by this report, we reviewed outcomes among patients with cGVHD at our center in an effort to validate these results and to assess the utility of early flare as a surrogate for survival in clinical trials.
Patient and Methods
We reviewed data collected in real time for 668 consecutive patients who were given initial systemic treatment for cGVHD diagnosed after May 1, 2001, following a first allogeneic HCT after myeloablative or nonmyeloablative conditioning regimens. Patients who had recurrent malignancy before the diagnosis of cGVHD were excluded. NIH criteria were used for the diagnosis of classic cGVHD, except that an ocular clinical score of 2 or 3 without Schirmer's test was accepted as sufficient to define ocular involvement [12]. The cohort did not include patients with persistent, recurrent, or delayed-onset acute GVHD (aGVHD) [12]. We used the first treatment change as a surrogate for exacerbation of cGVHD, and we evaluated the association of treatment change with NRM and overall survival (OS).
Change of Treatment
Dates and the reasons for change of treatment after the initial systemic treatment for cGVHD have been prospectively recorded at our center since May 2001. Change of treatment was defined as (1) addition of a new systemic medication to control cGVHD, (2) increase in the dose of prednisone to at least 1 mg/kg every other day for control of cGVHD, or (3) addition of a topical treatment for cGVHD involving a previously unaffected site. Change of treatment because of toxicity alone was not included, because the focus of the study was to use change of treatment as a surrogate for “flare” or inadequate control of cGVHD.
Statistical Methods
Results were analyzed according to information available as of May 2008. Cox proportional hazards regression was used to identify risk factors for treatment change, and time-dependent Cox proportional hazards regression was used to evaluate the association of treatment change with mortality. Cumulative incidence estimates and confidence intervals (CIs) were evaluated as previously described [13]. Factors evaluated for association with treatment change included age of the patient at time of transplant, recipient HLA-mismatching (ie, any graft-versus-host [GVH] mismatching at HLA-A, -B, -C, -DRB1, or -DQB1), donor type (ie, unrelated and related), recipient and donor sex, type of graft (ie, marrow or mobilized blood), type of preparative regimen (ie, nonmyelobative, myeloablative with and without total body irradiation), onset of cGVHD during systemic glucocorticoid treatment, type of onset of cGVHD (ie, progressive, de novo, quiescent), platelet count <100,000/μL at onset of cGVHD, and sites involved at the onset of cGVHD 7, 14.
Results
The median patient age at transplant of the cohort was 46.7 years (range: 0.5-73 years). Four hundred forty-six patients (67%) received a myeloablative conditioning regimen pretransplant. Other demographic and graft characteristics for the 668 patients are summarized in Table 1 and Figure 1. cGVHD was diagnosed during steroid treatment in 37% of the patients, and 30% had a platelet count <100,000/μL at the onset of cGVHD. Overall, 49% of patients had at least 1 of these risk factors as an indicator or poor prognosis. The distribution of organ involvement was typical for patients with cGVHD. Among the 668 patients, 56% had a treatment change within 2 years after beginning systemic treatment for cGVHD, and the median interval time from initial systemic treatment to the first treatment change was 4.4 months (range: 0.3-54 months) (Figure 2A). Treatment change was associated with 2 risk factors: skin involvement at the onset of cGVHD and use of a female donor for male recipients (Table 2). The cumulative incidence of NRM at 2 years was 16% (95% CI, 13%-19%) (Figure 2B), and OS at 2 years was 74% (95% CI, 70%-77%). As shown in Figure 3, the cumulative incidence of NRM was higher for patients who had a treatment change within 4 months after the initiation of systemic treatment for cGVHD than for those who did not (27% versus 12%, respectively).
Table 1. Characteristics of the Study Cohort (n = 668)
| Characteristic | N (%) |
|---|---|
| Demographic and graft characteristics | |
 Patient age ≥50 years | 271 (41) |
| Unrelated donor | 354 (53) |
| HLA-mismatched recipient | 105 (16) |
| Female donor, male recipient | 187 (28) |
| Mobilized blood cell graft | 573 (86) |
| Cord blood | 11 (2) |
| Nonmyeloablative conditioning regimens | 222 (33) |
| Myeloablative conditioning regimens: | |
| – with total body irradiation | 186 (28) |
| – without total body irradiation | 260 (39) |
| Chronic GVHD characteristics at onset | |
| Diagnosis during steroid treatment | 244 (37) |
| Progressive | 77 (12) |
| Quiescent | 438 (66) |
| De novo | 153 (23) |
| Platelet count <100,000/μL | 201 (30) |
Figure 1
Sites affected by cGVHD at time of diagnosis. Columns show the proportions of patients with organs affected by cGVHD at diagnosis.
Figure 2
(A) Cumulative incidence of first change in treatment after initial systemic treatment for cGVHD. (B) Cumulative incidence of NRM after the initial diagnosis of cGVHD.
Table 2. Risk Factors for Treatment Change
| Univariate | Multivariate | |||
|---|---|---|---|---|
| Characteristic | HR (95% CI) | P | HR (95% CI) | P |
| Female donor, male recipient | 1.31 (1.1-1.6) | .02 | 1.31 (1.0-1.6) | .01 |
| Skin involved at onset | 1.26 (1.0-1.6) | .04 | 1.26 (1.0-1.6) | .04 |
Figure 3
The cumulative incidence of NRM was higher for patients who had a treatment change within 4 months after initial systemic treatment for cGVHD (solid line) than for those who did not have a treatment change during the first 4 months (dashed line).
As expected, time-dependent analysis showed that treatment change after initial systemic treatment for cGVHD was associated with an increased risk of NRM (Table 3). Because the median time of treatment change was 4.3 months after initial systemic treatment for cGVHD, we evaluated results for patients who had a treatment change before 4 months compared to change after 4 months. As shown in Table 3, the association with NRM was stronger for early treatment change than for late change, although the difference between early and late change was not statistically significant (P = .06). A similar trend was found when we evaluated results for treatment change before 6 months compared to treatment change after 6 months (P = .09). When time of change was analyzed continuously, change at 4 months was associated with a 2.5-fold increase in NRM, and for each month of delay in treatment change, this effect on NRM was attenuated by a factor of 0.94, equivalent to a 6% reduction per month (Table 3). An association between treatment change and overall mortality was also noted, but the hazard ratios were lower (data not shown). For each month of delay in treatment change, the effect on overall mortality was attenuated by a factor of 0.93.
Table 3. Factors Associated with Nonrelapse Mortality
| Characteristic | HR (95% CI) | P |
|---|---|---|
| Treatment change at any time versus none | 2.53 (1.7-3.7) | <.0001 |
| Treatment change before 4 months versus none | 2.84 (1.9-4.2) | |
| Treatment change after 4 months versus none | 1.84 (1.1-3.1) | |
| Effect after versus before 4 months | .06 | |
| Treatment change before 6 months versus none | 2.72 (1.8-4.0) | |
| Treatment change after 6 monthes versus none | 1.77 (1.0-3.2) | |
| Effect after versus before 6 months | .09 | |
| Treatment change at 4 months versus none | 2.45 (1.7-3.6) | <.0001 |
| Relative effect per month | 0.94 (0.89-0.99) | .02 |
Discussion
Most phases II studies have relied on “response” as the primary endpoint in clinical trials for cGVHD. Although “response” is easily recognized in clinical practice, documentation of response for purposes of clinical trials (ie, convincing other people who cannot interview or examine the patient) is much more difficult and burdensome. Disease characteristics such as the severity of sclerosis or fasciitis and the intensity of cutaneous or mucosal erythema do not lend themselves to quantitative measurement, which makes it difficult to document the magnitude of change from baseline. The difficulty of documenting response made it attractive to consider change of treatment as an alternative endpoint for clinical trials, because this endpoint is objective and easy to document, although it measures failure rather than success and would be susceptible to bias in open-labeled studies.
Our results confirmed those reported by Kim et al. [11] in demonstrating that exacerbation of cGVHD is associated with an increased risk of NRM and with decreased OS. As in the Korean study, the associations with NRM and survival endpoints are stronger for early exacerbation than for late exacerbation. Skin involvement at onset of cGVHD and the use of female donor for male recipients were the only risk factors associated with treatment change. Because accrual of the cohort antedated the NIH consensus definition of severity scores for cGVHD [12], these factors could not be analyzed as risk factors for change of therapy and nonrelapse survival in our study.
We evaluated whether early treatment change could be used as a surrogate for NRM in clinical trials (see Supplemental Table). In our cohort, 27% of the patients had a treatment change within 4 months after diagnosis of cGVHD, and the risk of NRM at 2 years for these patients was 27%. The contribution to NRM at 2 years among these patients is 0.27 × 0.27, or 7%. Among the remaining 73% of patients who did not have a treatment change during the first 4 months, the risk of NRM was 12%, and the contribution to NRM at 2 years among these patients is 9%. The total NRM for both groups at 2 years is 16%. If we had a therapy that could eliminate the need for any treatment change during the first 4 months after diagnosis, the expected NRM at 2 years would be 12%, only 4 percentage points lower than historical results. Under the same assumptions, the difference in overall mortality would be only 2 percentage points. When the same analysis was repeated using 1 year (by which time 47% of patients had a treatment change) instead of 4 months as the time point for assessment of treatment change, the difference in NRM at 2 years was only 3 percentage points.
Our results indicate that differences in the strength of associations between early change of treatment and NRM or survival are not large enough to allow the use of early exacerbation as a surrogate endpoint for NRM or survival in clinical trials for treatment of cGVHD. In evaluating our results, it should be noted that the acuity of cGVHD was higher in the Korean cohort [11] than in our cohort, as indicated by a higher incidence of progressive onset (38% versus 12%) and thrombocytopenia (49% versus 30%), a higher incidence of exacerbation at 2 years (70% versus 56%), a shorter time to exacerbation, (2.4 versus 4.4 months), and a higher incidence of NRM at 2 years (39% versus 16%), respectively. In addition, the previous report included patients with late, persistent, or recurrent aGVHD, whereas our cohort included only patients with classic cGVHD and overlap syndrome [12]. These differences might account for the stronger association between early change of treatment and NRM in the Korean study than in our study. We conclude that other measures of clinical benefit, such as response, will need to be developed as endpoints in phase II trials of treatment for cGVHD.
Acknowledgments
The authors thank the staff of the Long-Term Follow-Up for invaluable assistance with data collection including Judy Campbell, Colleen McKinnon, Catherine Baker, Gina Brooks and Linda Guerrero. They are also very grateful to our patients for their participation in clinical trials and the referring physicians and medical staff for their collaborative efforts in the excellent care provided to patients and families.
Financial disclosure: This work was supported in part by grants CA 118953-01A1 and CA78902 from the National Institutes of Health, Department of Health and Human Services.
Authorship
Contribution: Mary E.D. Flowers: study design, analyzed data, and wrote the paper; Paul Martin: analyzed the data and wrote the paper. Barry Storer performed the statistical analysis, Paul Carpenter, Afonso Vigorito, Paulo Campregher, Andrew Rezvani, Carina Moravec, Hans-Peter Kiem, Fero Matthew, George Georges, Edus Warren, Stephanie Lee, and Jean Sanders: collected the data for the study. Fred Appelbaum: helped to write the paper.
References
- Health and functional status of long-term survivors of bone marrow transplantation. EBMT Working Party on Late Effects and EULEP Study Group on Late Effects. Ann Intern Med. 1997;126:184–192
- Long-term survival and late deaths after allogeneic bone marrow transplantation. Late Effects Working Committee of the International Bone Marrow Transplant Registry. N Engl J Med. 1999;341:14–21
- . Recovery after allogeneic marrow transplantation: prospective study of predictors of long-term physical and psychosocial functioning. Bone Marrow Transplant. 1993;11:319–327
- Morbidity and mortality of chronic GVHD after hematopoietic stem cell transplantation from HLA-identical siblings for patients with aplastic or refractory anemias. Biol Blood Marrow Transplant. 2002;8:47–56
- . Chronic graft-versus-host disease. Biol Blood Marrow Transplant. 2003;9:215–233
- Immune reconstitution after allogeneic marrow transplantation compared with blood stem cell transplantation. Blood. 2001;97:3380–3389
- Duration of immunosuppressive treatment for chronic graft-versus-host disease. Blood. 2004;104:3501–3506
- . Study design and endpoints in graft-versus-host disease. Best Practice Res Clin Haematol. 2008;21:357–372
- National Institutes of Health consensus development project on criteria for clinical trials in chronic graft-versus-host disease: VI. Design of Clinical Trials Working Group report. Biol Blood Marrow Transplant. 2006;12:491–505
- Measuring therapeutic response in chronic graft-versus-host disease: National Institutes of Health consensus development project on criteria for clinical trials in chronic graft-versus-host disease: IV. Response Criteria Working Group report. Biol Blood Marrow Transplant. 2006;12:252–266
- Time to first flare-up episode of GVHD can stratify patients according to their prognosis during clinical course of progressive- or quiescent-type chronic GVHD. Bone Marrow Transplant. 2007;40:779–784
- National Institutes of Health consensus development project on criteria for clinical trials in chronic graft-versus-host disease: I. Diagnosis and Staging Working Group report. Biol Blood Marrow Transplant. 2005;11:945–956
- . Estimation of failure probabilities in the presence of competing risks: new representations of old estimators. Stat Med. 1999;18:695–706
- Graft-versus-host disease (GVHD)-specific survival and duration of systemic immunosuppressive treatment in patients who developed chronic GVHD following allogeneic haematopoietic cell transplantation. Br J Haematol. 2003;122:637–644
Financial disclosure: See Acknowledgments on page 1383.
PII: S1083-8791(08)00416-3
doi:10.1016/j.bbmt.2008.09.017
© 2008 American Society for Blood and Marrow Transplantation. Published by Elsevier Inc. All rights reserved.
Volume 14, Issue 12 , Pages 1380-1384, December 2008
