Association of Human Development Index with rates and outcomes of hematopoietic stem cell transplantation for patients with acute leukemia

Hematopoietic stem cell transplantation (HSCT) is a well-recognized method with documented curative potential for the treatment of patients with both acute myeloid leukemia (AML) and acute lymphoblastic leukemia (ALL).^1,2  Therapeutic options include allogeneic (allo-) and autologous (auto-) HSCT. Allo-HSCT may be performed from either human leukocyte antigen (HLA)-matched or mismatched, related or unrelated donors. HSCT is an expensive procedure, requiring well-organized and equipped transplantation units, advanced supportive care, and highly qualified medical and nursing staff. As life-threatening complications may appear late, adequate monitoring and treatment are necessary for many months to years after transplantation.³  Therefore, it may be expected that the access to HSCT as well as outcome of the procedure may depend on socioeconomic factors varying among and within countries. Identification of these factors may be essential to rationalize national therapeutic strategies for patients with acute leukemia.

Several studies performed in either developing countries, such as Brazil, or those with high ethnic and demographic diversities, such as the United States or the United Kingdom, demonstrated the association of the socioeconomic status (SES) with the access to HSCT and mortality after transplantation.^4-9  In Europe, as demonstrated by Gratwohl et al, the distribution of transplantations varies strongly among countries, which, in major part, depends on economic conditions defined by the Gross National Income per capita.^10,11  On the other hand, it may be speculated that the background determining availability of HSCT may be more complex and may include socioeconomic factors other than Gross National Income. Furthermore, the impact of the SES on the results of transplantation has not been studied so far in Europe.

The Human Development Index (HDI) is commissioned by the United Nations Organization to evaluate a country's socioeconomic achievements in 3 basic aspects: longevity, knowledge, and standard of living.¹²  Longevity is measured by life expectancy at birth and expressed as Life Expectancy Index (LEI). Knowledge is evaluated by a combination of adult literacy rate and the rate of enrollment to primary, secondary, and tertiary schools (Education Index [EI]). Standard of living is measured by the Gross Domestic Product per capita, purchasing power parity US dollars (Gross Domestic Product Index [GDPI]). According to the HDI, the countries may have a low (HDI < 0.500), intermediate (HDI = 0.500-0.799), or high (HDI ≥ 0.800) development status. Most of the European states belong to the latter category, but even then, there are variations. In this study, we hypothesized that all components used to calculate the HDI may contribute to the availability of HSCT and the treatment results. Hence, our goal was to evaluate the association between the HDI, LEI, EI, and GDPI, and the number of transplantations performed per population as well as early and long-term outcome of adult patients with acute de novo leukemia referred for HSCT.

This was a retrospective multicenter analysis. Data were provided by the registry of the Acute Leukemia Working Party of the European Group for Blood and Marrow Transplantation (EBMT). Centers participating in the EBMT are annually requested to report all consecutive stem cell transplantations and follow-up. The validation and quality control program includes verification of the computer printout of the entered data, cross-checking with the national registries, and on-site visits of selected teams.

For the analysis of transplantation rates, inclusion criteria were as follows: (1) diagnosis of AML or ALL, (2) age more than or equal to 18 years, and (3) HSCT performed between 2001 and 2005 in one of 30 European countries (including Israel), reporting to the EBMT. Altogether, 16 403 patients treated with autologous (n = 4641), syngeneic (n = 69), or allogeneic (n = 11 693) HSCT were included.

The impact of socioeconomic factors on outcome, besides fulfilling the aforementioned criteria, was restricted to: (1) patients with de novo AML in first complete remission, (2) HSCT from HLA-identical sibling or from HLA-matched unrelated donor (the minimum donor-recipient compatibility criterion was based on HLA-A, -B, and –DR matching using serologic or low resolution molecular typing), (3) HSCT using bone marrow or peripheral blood as a source of stem cells (cord blood transplantations were excluded), (4) HSCTs with myeloablative conditioning (definition of myeloablative regimen was considered if busulfan was administered at the total dose of ≥ 8 mg/kg or total body irradiation was applied at ≥ 6 Gy), and (5) HSCT without ex vivo T-depletion. For this part of the study, data on 2015 patients were analyzed.

Detailed characteristics of patients, donors, and HSCT procedure are listed in Table 1.

Data on the HDI, LEI, EI, and GDPI for European countries in 2001, 2002, 2003, 2004, and 2005 were collected from the Web site of the United Nations Organization.^13-17  For particular indices, mean values in the 5-year period were calculated. For the analysis of outcome, countries were ranked and categorized by quintiles to divide the whole group in cohorts with equivalent number of HSCT procedures. Mean values and results of categorization are presented in Table 2.

Correlations between LEI, EI, and GDPI and transplantation rates calculated per 1 million inhabitants were studied using the Pearson correlation coefficient. In the analysis of outcome, the probabilities of leukemia-free survival (LFS), relapse incidence (RI), and nonrelapse mortality (NRM) were the primary study endpoints. The LFS was defined as time interval from MRD-HSCT to either relapse or death in remission and was calculated using the Kaplan-Meier estimate. The RI and NRM were calculated using cumulative incidence curves in a competing risks setting, with death in remission being treated as a competing event to relapse.^18,19  Univariate analyses were done with the use of log-rank test for LFS, whereas Gray test was applied for RI and NRM.

For the analysis of outcome, all countries were first categorized using the quintiles with increasing values of the SES indices. If the relative event rates (ratio of the observed number of events to the expected number of events in a category with the assumption of no variation across categories) in 2 or more adjacent categories were not substantially different, the categories were merged. If a linear trend was observed in the relative event rates, the variable was used as a continuous factor. Regarding HDI, only the fifth group (the highest HDI) differed from the other 4 categories, which therefore were merged. Multivariate analysis was performed with use of the Cox proportional hazard model, adjusted for the donor type (sibing vs unrelated), recipient age (less than vs more than or equal to median), year of transplantation (before 2003 vs 2003 or later), type of conditioning (total body irradiation-based vs chemotherapy-based), the interval from diagnosis to HSCT (less than vs more than or equal to median), and the source of stem cells (peripheral blood vs bone marrow). The median follow-up for survivors was 36 months (range, 0.1-91 months).

All P values are 2-sided with type 1 error rate fixed at .05. Statistical analyses were performed with SPSS Version 15.0 and S-Plus Version 6.1 (MathSoft Inc) software packages.

The total number of transplantations calculated per 1 million inhabitants correlated with the HDI (correlation coefficient, R = 0.78; P < .001). When subindices were analyzed separately, the strongest associations were found for HSCT rates with LEI (R = 0.84, P < .001) and GDPI (0.77, P < .001), and only marginal significance was observed in case of EI (R = 0.39, P = .03; Table 3). Similar patterns of correlations were observed in an analysis restricted to sibling HSCT and unrelated HSCT. For the rates of auto-HSCT, the correlation coefficients with the HDI, LEI, and GDPI were lower, but statistically significant, whereas no correlation was found with the EI (Table 3).

The association of the SES with outcome was analyzed for patients with AML treated with HLA-matched related or unrelated HSCT in first complete remission. The probabilities of LFS at 3 years for the 5 consecutive groups of countries with increasing HDI were as follows: 56% plus or minus 3%, 59% plus or minus 2%, 63% plus or minus 3%, 58% plus or minus 4%, and 68% plus or minus 3% (P = .01; Table 4; Figure 1). In a multivariate analysis, transplantations performed in countries belonging to the upper HDI category were associated with significantly higher LFS compared with the remaining ones (HR = 1.36, P = .008; Table 5). In a univariate analysis, significant differences were also found for the categories of countries classified according to the GDPI (Table 4). The LFS rates with increasing GDPI were 57% plus or minus 3%, 54% plus or minus 3%, 66% plus or minus 3%, 65% plus or minus 2%, and 63% plus or minus 3% (P = .001). However, the independent impact of the GDPI status on the probability of LFS has not been confirmed in a multivariate model (P = .09; Table 5). Similarly, no significant associations were found for LFS in relation to LEI and EI.

The cumulative incidence of early transplantation-related mortality (up to day 100) for the 5 consecutive groups of countries with increasing HDI were as follows: 7% plus or minus 1%, 6% plus or minus 1%, 9% plus or minus 1%, 8% plus or minus 2%, and 5% plus or minus 1% (P = .74; Table 4). Accordingly, the overall NRM at 3 years equaled 22% plus or minus 3%, 19% plus or minus 2%, 18% plus or minus 2%, 20% plus or minus 3%, and 16% plus or minus 2% (P = .2; Table 4; Figure 1). In a multivariate model adjusted to other potential risk factors, there was a tendency for decreased risk of NRM in countries with the highest HDI; however, the effect did not reach statistical significance (P = .1; Table 5). As well, the impact on NRM could not be demonstrated for any of the analyzed subindices (Table 5).

The cumulative incidence of relapse with significant difference for the upper HDI category compared with the remaining ones (P = .03; Table 4; Figure 1). In the Cox model, the mean HDI more than 0.9238 independently reduced the risk of relapse (hazard ratio, HR = 0.72; P = .04; Table 5). With regard to subindices contributing to the HDI, the univariate analysis revealed a lower incidence of relapse for the 3 upper classes as defined by the GDPI (a range of 21% ± 2% to 22% ± 3%) compared with the 2 lower GDPI categories (29% ± 3% and 32% ± 3%; Table 4). However, after adjustment to other prognostic factors, the effect of the GDPI more than 0.9238 lost its significance (P = .16; Table 5). The LEI and EI did not appear to be associated with the incidence of relapse.

The impact of socioeconomic factors on the treatment of patents with acute leukemia was studied by several investigators. Factors, such as the parental educational level, father's occupation, ethnicity, and SES-related malnutrition, were found to be associated with survival of pediatric ALL patients in the Brazilian, North American, Australian, and Dutch populations.^20-27  With regard to adults with acute leukemia, the impact of the SES on survival was indirectly demonstrated by studies comparing patients treated between 1982 and 1996 in Estonia and Western Sweden, the 2 regions characterized by different political and economic systems, with similar ethnic and geographic background.^28,29  Swedish patients had a significantly higher chance of cure compared with their Estonian counterparts.^28,29  Fagundes et al, using the HDI to evaluate the SES of adults with AML in Brazil, showed the differences of outcome, which were in major part dependent on the compliance to chemotherapy protocol.³⁰ 

Until now, few studies focused on the association of socioeconomic factors with HSCT. The analysis by Mitchell et al revealed that, in the United States, between 1988 and 1991, the access to HSCT was restricted for black compared with white patients, which resulted mainly from the differences in the insurance coverage.⁹  In 2 subsequent studies, it was demonstrated that the American Hispanic patients with either acute or chronic leukemias treated with HSCT had lower survival rates compared with the whites, raising the need to further examine the role of social, economic, and cultural factors.^4,5  Indeed, Armenian et al showed that Hispanic Americans are less likely to self-report the presence of life-threatening chronic post-transplantation complications compared with non-Hispanic Americans, which was explained mainly by the differences in the educational status.⁸  Similarly to the American studies, an adverse effect of ethnicity (black vs white patients) with regard to NRM was reported in the United Kingdom.⁶  Finally, the negative impact of poor SES on outcome of HSCT has recently been demonstrated for patients treated in Brazil.⁷ 

In view of the so far published studies, it appears that both the availability and outcome of HSCT may depend on many factors related to the SES of patients with acute leukemia. In Europe, the social, economic, cultural, and ethnic diversity is less pronounced compared with other regions of the world, and the majority of countries are classified as well developed. Even then, however, the differences exist with regard to the standard of living, education, and organization of the healthcare, which may be reflected by variations in the GDPI, EI, LEI, and the composite HDI, all used in our study as surrogates of the SES.

As HSCT, especially allo-HSCT, is an expensive way of treatment, it could be thought that the access to this procedure is restricted mainly by the financial coverage available in particular countries, dependent on the national income. Indeed, in line with the finding of Gratwohl et al,^10,11  we demonstrated a strong correlation of HSCT rates with the GDPI. However, the effect was even stronger with regard to the LEI and could also be demonstrated for the EI. Hence, not only purely economic conditions, but probably also organization of the healthcare and education system, translates into the availability of HSCT. In addition, allo-HSCT is a very complex procedure that involves several professional skills (eg, hematology, internal medicine, radiotherapy, sophisticated laboratories), which probably depend on the HDI. As expected, the correlations of the HDI and subindices with the transplantation rates were more pronounced in the case of allo-HSCT, compared with auto-HSCT, reflecting substantial differences associated with the logistics and costs of the aforementioned procedures. Our analysis was based on the assumption that the incidence of acute leukemia does not vary substantially among countries and does not depend on the HDI. Although exact epidemiologic data from all countries were unavailable, some publications suggest no association of the SES with rates of leukemia in Europe.³¹ 

Results of our study indicate that in Europe the HDI is associated not only with the availability but also with results of transplantation. Studying HLA-matched, T-replete, myeloablative allo-HSCT for AML patients in first complete remission as a representative model with well-established indications and a quite homogeneous population, we showed that the probability of LFS was significantly increased for patients treated in 8 countries with the highest HDI, compared with the remaining 22 states where the results were super-imposable. Hence, our findings indicate that, apart from the transplantation rates, European countries with lower HDI (first quintile) may achieve comparable quality of treatment to those better-developed (second to fourth HDI quintile). On the other hand, it appears that, for the majority of European states, there is still room for improvement to reach the outcomes achieved by the highest-HDI states. Identification of specific factors contributing to these differences seems essential.

Allo-HSCT raises a lot of requirements related to the adequate donor selection, standard of hospitalization, intensive and expensive supportive therapy, and highly specialized medical monitoring for a long period after transplantation. The reasons of failure are leukemia relapse as well as early and late complications, including graft-versus-host disease and infections. Whereas early mortality may vary according to the standard of hospital care, late events depend in major part on the patients' behavior during their stay at home. Hence, exposure to infectious agents, awareness of possible complications, and easy access to specialized medical staff are all of importance. Our analysis showed only a tendency for the association of the HDI with the overall NRM. No difference could be demonstrated at day 100. Based on these findings, we can speculate that the quality of medical care during the hospital stay is equivalent across Europe.

In contrast to the NRM, the effect of HDI on the risk of relapse could be demonstrated in both univariate and multivariate analyses, with decreased RI in countries with the highest HDI. The risk of relapse may depend on the disease biologic features, quality of preceding induction-consolidation chemotherapy, and the transplantation procedure itself. To explore the possible reasons of our findings, we compared the patient and HSCT characteristics for countries belonging to the upper HDI category with the remaining ones (Table 6). We found that better results in countries with the highest HDI were achieved despite increased recipient age. Similarly, the interval from diagnosis to HSCT was decreased in these countries, whereas in the Cox model prolonged interval diagnosis HSCT was associated with reduced risk of relapse. The fact that results in countries with the highest HDI are improved across this factor may indicate higher efficacy of the induction-consolidation regimens or better compliance to chemotherapy protocol, in particular higher density of chemotherapy courses. The intensity of initial chemotherapy and the duration of the total induction and consolidation program may influence the quality of remission at the time of allo-HSCT and translate into improved long-term outcome. Finally, total body irradiation-based conditioning was more frequently used in countries with the highest HDI compared with the remaining ones. However, because the advantageous effect of total body irradiation has been independently confirmed in a multivariate model, it cannot explain the impact of HDI on RI.

Trying to identify specific aspects of the SES that contribute to the effect of HDI on results of allo-HSCT, we analyzed separately the potential influence of the LEI, EI, and GDPI. Although results of a univariate analysis suggested the association of GDPI with all LFS, RI, and NRM, results of the multivariate analysis did not confirm it. As well, no significant effect could be demonstrated for the LEI and EI. Therefore, it appears that the impact of SES on results of allo-HSCT is multifactorial and that the HDI as a composite index may be representative.

Altogether, results of our study indicate that the HDI, being a surrogate of the SES, is associated with the rates of all types of HSCT in Europe, as well as the outcome of patients with AML treated with allo-HSCT. However, the positive effect on results is observed only with regard to few countries with particularly high HDI, whereas the outcome in the remaining ones is comparable. Some limitations of our study should, however, be emphasized. Because of its retrospective nature, some important information was lacking, including data on cytogenetic and molecular risk factors, details on induction, and consolidation protocols, reasons of NRM. Therefore, there is need for further prospective studies with detailed patient and procedure characteristics. In addition, future investigation should take into account particular aspects of the SES, including organization of the health system, allocation of transplantation centers, as well as environmental factors, which differ among countries and regions. Identification of detailed relationships between socioeconomic factors and results of allo-HSCT may indicate areas for future improvement.

Contribution: S.G. designed and performed research, interpreted data, and prepared the manuscript; M.L. analyzed data and interpreted results; G.E., D. Beelen, D. Blaise, A.G., A.B., T.C., and J.H. collected and provided data; E.M.F., E.N., and F.F. interpreted data and reviewed manuscript; and V.R. designed research, interpreted data, and reviewed manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Sebastian Giebel, Department of Clinical and Experimental Oncology, Comprehensive Cancer Centre, Maria Sklodowska-Curie Memorial Institute Branch Gliwice, Wybrzeze Armii Krajowej 15, 44–101 Gliwice, Poland; e-mail: sgiebel@poczta.onet.pl.