Introduction

The biological and clinical features of T- and B-lineage adult acute lymphoblastic leukemia (ALL) are considered to be different [1–3]. In fact, several clinical studies have demonstrated this fact based on different results in T- and B-lineage ALL [4–12].

Intensive combination chemotherapy regimens have improved the outcome of pediatric patients with ALL [13]. Based on this success, intensive regimens have been adopted for the treatment of adult patients with ALL, and these have led to better outcomes [14–16]. With regard to pediatric regimens, clinical studies have examined the dose intensification of non-myelosuppressive drugs. A superior outcome is achieved with the use of dexamethasone, compared to predonisone [17, 18], and high-dose methotrexate has also been shown to be effective [19]. In addition, the intensified use of l-asparaginase could improve the outcome [20]. Moreover, studies in pediatric patients have suggested that the dose intensification of drugs, such as methotrexate and l-asparaginase, may have different effects in T- and B-lineage ALL [21, 22]. However, the role of each individual drug has not been well established in adult regimens.

Therefore, in the present study, we performed meta-analysis and meta-regression analysis to compare the outcomes of chemotherapy for T- and B-lineage ALL.

Methods

Study selection

In April 2013, we performed a literature search for clinical studies published in English from January 1998 to March 2013 through PubMed by using key words, “acute lymphoblastic leukemia OR acute lymphocytic leukemia OR acute lymphoid leukemia AND adult AND chemotherapy”. In addition, the studies were restricted to those in humans. The titles and abstracts of the identified articles were reviewed. Studies were eligible for inclusion in the analyses if they included more than 80 patients with adult ALL who received the primary treatment with a uniform regimen, and if a hazard ratio (HR) and its 95 % confidence interval (CI) comparing overall survival (OS), disease-free survival (DFS), or event-free survival (EFS) between patients with T- and B-lineage ALL were either reported or could be calculated. Studies that included only adolescent or elderly patients were excluded.

Outcome measures and data extraction

The primary outcome measure was OS. However, if OS was not reported in the identified studies, DFS or EFS was substituted for OS. Characteristics that were extracted from each study were the name of the first author, study title, year of publication, numbers of total patients, patients with T-lineage ALL, and those with B-lineage ALL, age of included patients, inclusion of Philadelphia chromosome-positive (Ph+) ALL, the rate of achieving first complete remission (CR1), the rate of transplantation in CR1, and HR and its 95 % CI for the primary outcome measure. In addition, the use of dexamethasone (Dex), total dose of cytosine arabinoside (AraC), total dose of methotrexate (MTX), and total dose of l-asparaginase (l-asp) during induction and consolidation regimens was collected. The dose of l-asp was calculated according to the International Union of Biochemistry (I.U.), and one unit or one Kyowa Unit (K.U.) was considered to be equal to 0.86 I.U.

If the HR and/or its 95 % CI of the primary outcome measure could not be extracted directly, the numbers of patients with T-lineage and those with B-lineage ALL, the total number of events of the primary outcome measure, and the P value were extracted. The HR and its 95 % CI were then calculated from these data [23].

Statistical analyses

We used the fixed-effect model and random-effect model to combine the HR in each study. Interstudy heterogeneity was tested by using the Cochran Q statistic (chi-square value), with a P value of less than 0.1 considered to be significant, and was quantified by using the I 2 statistic, where a value of 50 % or greater indicates substantial heterogeneity. Meta-regression analyses were performed using the factors extracted from each study, and factors with a P value of equal to or less than 0.1 were considered to be significant. We also performed a multivariate meta-regression analysis using significant factors. All statistical analyses were performed with EZR (Saitama Medical Center, Jichi Medical University) [24], which is a graphical user interface for R (The R Foundation for Statistical Computing, version 2.13.0, Vienna, Austria). It uses a “meta” package for meta-analysis and a “metatest” package for meta-regression analysis [25].

Results

The primary literature search yielded 2870 articles, most of which could be excluded by screening their titles. The abstracts of the remaining 134 articles were reviewed. Finally, 11 clinical studies that were published between 1998 and 2013 were included in the following meta-analysis [7–10, 14, 16, 26–30].

The study characteristics are summarized in Table 1. These studies included a total of 381 patients with T-lineage ALL and 1366 patients with B-lineage ALL. Nine studies included patients with Ph+ ALL, but no study used tyrosine kinase inhibitors (TKIs) for those with Ph+ ALL. The primary outcome measure was OS in seven studies, DFS in two, and EFS in two.

Table 1 Study characteristics
Full size table

Meta-analyses using the fixed-effect model and the random-effect model demonstrated superior survival in patients with T-lineage ALL compared to those with B-lineage ALL (HR 1.88, 95 % CI 1.73–2.05, and HR 1.78, 95 % CI 1.50–2.11, respectively), though there was significant heterogeneity among studies (P < 0.0001) (Fig 1). We thought that this heterogeneity was attributed to whether or not patients with Ph+ ALL were included in each study, and therefore, we performed a meta-regression analysis regarding Ph+ ALL. The inclusion of patients with Ph+ ALL in a study showed a significant and positive correlation with superior survival in patients with T-lineage ALL (coefficient 0.49, P = 0.007), although superior survival in patients with T-lineage ALL was preserved at least in the fixed-effect model, even if studies that did not include patients with Ph+ ALL were separately analyzed (HR 1.28, 95 % CI 1.03–1.60 in the fixed-effect model and HR 1.14, 95 % CI 0.66–1.97 in the random-effect model). Next, we performed meta-regression analyses, adjusted for whether or not patients with Ph+ ALL were included in each study. Use of Dex, higher total dose of MTX, and higher total dose of l-asp in induction and consolidation regimens were associated with a significant trend toward a better outcome in T-lineage ALL compared to that in B-lineage ALL (Table 2). A meta-regression analysis including Dex and the dose of l-asp together as covariates showed that the effects of these factors were independently significant (P = 0.023 and P = 0.002, respectively) (Table 3a), and a meta-regression analysis including Dex and the dose of MTX together as covariates also showed that the effects of these factors were independently significant (P = 0.102 and P = 0.003, respectively) (Table 3b). We could not compare the effects of the doses of MTX and l-asp to avoid multi-collinearity, since there was a strong correlation between these doses (data not shown).

Fig. 1
figure 1

Forest plot of the hazard ratio (HR) and 95 % confidence interval (CI) for survival in each study and the entire study. Meta-analyses using the fixed-effect model and the random-effect model demonstrated superior survival in patients with T-lineage acute lymphoblastic leukemia (ALL) compared to those with B-lineage ALL, though there was significant heterogeneity

Full size image
Table 2 Meta-regression analysis adjusted for whether or not patients with Philadelphia chromosome-positive acute lymphoblastic leukemia were included
Full size table
Table 3 Multivariate meta-regression analysis including significant factors
Full size table

Discussion

Many clinical studies have compared T- and B-lineage ALL [4–12]. However, the results of these comparisons have been inconsistent, partially because the studies included only a limited number of patients with T-lineage ALL, due to the relative rarity of T-lineage ALL in adult patients. Therefore, we performed a meta-analysis that included 381 patients with T-lineage ALL and found a superior outcome for T-lineage ALL.

We have to be careful about the significant interstudy heterogeneity in our meta-analysis. This heterogeneity might be attributed to the difference regarding whether or not each study included patients with Ph+ ALL. The inclusion of patients with Ph+ ALL in a study was positively related to superior survival in patients with T-lineage ALL. Although the inclusion of tyrosine kinase inhibitors (TKIs) in the present treatment of patients with Ph+ ALL has dramatically improved the outcomes of these patients [31], TKIs were not used in patients with Ph+ ALL in the studies that we included, since we chose clinical studies, in which uniform regimens were used in patients with T- and B-lineage ALL and we could not obtain the clinical data regarding patients limited to those with Ph-negative ALL from published articles. This fact must have influenced the results of the meta-analysis strongly. Superior survival in patients with T-lineage ALL was preserved at least in a fixed-effect model even if studies that did not include patients with Ph+ ALL were analyzed separately.

In addition, we should be cautious, since we only included studies that made it possible to determine a HR and its 95 % CI for the comparison of survival between patients with T- and B-lineage ALL. It is possible that the authors who found a difference between T- and B-lineage ALL were more likely to provide detailed results, and this could have led to a reporting bias.

Through our meta-regression analyses, we demonstrated that the use of Dex, a higher total dose of MTX, or a higher total dose of l-asp might improve the outcome of adult patients with T-lineage ALL. Dose intensification of non-myelosuppressive drugs, such as MTX and l-asp, has been used in pediatric regimens and has led to excellent results, especially in pediatric patients with T-lineage ALL [21, 22]. The efficacy of dose intensification of non-myelosuppressive drugs, especially for T-lineage ALL, seems to be a common feature, independent of patient’s age, although we have to confirm this result through clinical data regarding patients limited to those with Ph-negative ALL. Intensified chemotherapy based on pediatric regimens is now being extended to adolescents and young adults [32], and even to older patients in recent trials [16], and this treatment strategy might be considered more favorably for T-lineage ALL, even in adult patients.

In conclusion, the use of Dex and high-dose l-asp or MTX may improve the outcome of T-lineage ALL. This hypothesis should be tested in a prospective study including only patients with Ph-negative ALL.