P-glycoprotein and BCL-2 levels predict outcome in adult acute lymphoblastic leukaemia


Maria Ilaria Del Principe, Cattedra Ematologia, Ospedale S Eugenio, Università‘Tor Vergata, P. le Umanesimo, 10, 00144 Roma, Italy. E-mail: midelprincipe@tiscalinet.it or g.delpoeta@tin.it


Summary. Concurrent resistance mechanisms, such as P-glycoprotein (PGP) and bcl-2, may contribute to a worse outcome in adult acute lymphoblastic leukaemia (ALL). Between 1990 and 2000, we analysed PGP and bcl-2 by flow cytometry, using two anti-PGP (C219 and JSB-1) monoclonal antibodies (mAbs) and an anti-bcl-2 mAb in 115 de novo adult ALL patients. Both a longer overall survival (OS) and longer disease-free survival (DFS) were observed in PGP-negative patients (23%vs 0% at 3 years, P = 0·011 and 29%vs 0% at 2 years, P = 0·006 for C219 respectively; 42%vs 0% at 1·5 years, P = 0·004 and 53%vs 0% at 8·5 months, P = 0·00006 for JSB-1 respectively). Bcl-2 positivity was associated with a significantly higher complete remission rate (90%vs 66%, P = 0·01). Moreover, in 69 patients not presenting with either t(9;22) or B-mature immunophenotype, PGP negativity (JSB-1) maintained its significant favourable prognostic impact with regard to OS (41%vs 0% at 1·5 years, P = 0·009) and DFS (83%vs 0% at 6 months, P = 0·0005). Importantly, within a subset of 62 patients with normal (n = 31) or unknown (n = 31) karyotype, PGP (JSB-1)-negative patients showed both a significantly longer OS and DFS (63%vs 0% at 1·4 years, P = 0·018 and 84%vs 0% at 6 months, P = 0·001 respectively). In multivariate analysis, JSB-1 (P = 0·008) and cytogenetics (P = 0·02) were found to be independent prognostic factors with regard to DFS. Therefore, in adult ALL, PGP and bcl-2 represent sensitive indicators of clinical outcome, and potential targets of novel molecules aimed at overcoming chemoresistance and recurrent relapses.

Several studies have investigated the expression of the multiple drug resistance 1 (MDR1) gene in leukaemic cells from acute lymphoblastic leukaemia (ALL) patients, either by using different probes (Rothenberg et al, 1989) or by studying the increase in MDR-1 mRNA and the over-expression of P-glycoprotein (PGP) (Marie et al, 1991), in an attempt to demonstrate a correlation with treatment response or patient follow-up. Although in acute myeloid leukaemia (AML) the prognostic role of PGP has been clearly established (Campos et al, 1992; Del Poeta et al, 1996), the role of PGP in ALL drug resistance is still currently debated (Goasguen et al, 1996; Nussler et al, 1996; Dhooge et al, 1999; Kanerva et al, 2001). Defects in the mitochondrial-mediated pathway of apoptosis may induce multidrug resistance; therefore, alterations in the expression of bcl-2 and related proteins, such as bax, can have a marked influence on chemosensitivity (Reed et al, 1996). High levels of bcl-2 expression have generally been shown to correlate with a poor treatment outcome in some haematological malignancies, including follicular lymphoma (Hermine et al, 1996), chronic lymphocytic leukaemia (Robertson et al, 1996) and AML (Campos et al, 1993). In contrast, other studies have indicated that increased levels of bcl-2 are not prognostic, or even associated with an improved event-free survival, in ALL (Coustan-Smith et al, 1996; Schimmer et al, 2001). In addition, it has been demonstrated that bcl-2 and bax play key roles in the control of proliferation as well as apoptosis. In this study, we proposed a reliable and feasible flow cytometric method for the quantification of PGP and bcl-2 expression levels, and demonstrated the significant prognostic value of PGP and bcl-2 measurements in adult ALL.

Patients and methods

Patients.  We studied 115 newly diagnosed adult ALL patients, admitted to our unit, between 1990 and 2000, as shown in Table I. The patients had a median age of 40 years (range 15–75 years) with a performance status ≤ 2 by the World Health Organization (WHO) grading system. There were six early pre-B-cell ALL (pre-B ALL), 39 common ALL, 18 pre-B ALL, 18 surface membrane immunoglobulin-positive ALL (SmIg+ ALL) and 34 T-cell ALL (T-ALL) patients. Seventy-two adult patients were treated with a conventional induction regimen [ALL 0288 Gruppo Italiano Malattie Ematologiche Maligne dell'Adulto (GIMEMA) protocol] based on the combination of vincristine, daunorubicin and prednisone with or without cyclophosphamide (Annino et al, 2002). Relapses that occurred after this protocol were treated with idarubicin and cytosine arabinoside at conventional doses. Forty-three patients underwent intensive induction chemotherapy based on high doses of cytarabine (3 g/m2 for 5 d) and mitoxantrone (80 mg/m2 for 1 d) plus prednisone (40 mg/m2 for 28 d), which, after a similar consolidation cycle, was followed by an autologous or allogeneic transplantation (Del Poeta et al, 1999). Thirty-two patients were transplanted and censored at the time of bone marrow (BM) or peripheral blood stem cell infusion. Relapsed patients were treated with allogeneic transplants whenever eligible or, alternatively, with fludarabine–cytosine arabinoside-based regimens. Finally, no significant difference in terms of the induction of complete remission (CR; P = 0·4), overall survival (OS; P = 0·5) and disease-free survival (DFS; P = 0·8) was observed between adult ALL patients treated with the ALL 0288 protocol and intensive chemotherapy followed by transplantation procedures.

Table I.   The clinical characteristics of adult ALL patients.
  • *

    Relapse and survival parameters are expressed in months.

Total number115
WBC (× 109/l)
 > 5076
 > 10010
Karyotype (n = 84)
Other abnormal karyotype22
 C219+ (%)43·7
 JSB-1+ (%)31·4
 Bcl-2+ (%)53·3
CR rate (%)82·1
Median time to relapse*5·6
Median duration of overall survival (OS)*18·2 20% at 11 years
Median duration of disease-free survival (DFS)*18·7 25% at 11 years

Diagnosis.  The diagnosis was established by cytological examination of BM smears according to the French–American–British (FAB) group recommendations (Bennett et al, 1976) and immunophenotyping of leukaemic cells (Sobol et al, 1985).

Immunophenotyping.  A multiparameter analysis based on double-colour immunofluorescence study of surface antigens was performed using combinations of phycoerythrin (PE)- and fluorescein isothyocyanate (FITC)-conjugated monoclonal antibodies (mAbs), as previously described (Del Poeta et al, 1994). The results were obtained by gating the appropriate blast gate using side scatter (SSC) versus forward light scatter (FSC), and examining the associated FITC and PE expression. We used mAbs against CD2, CD3, CD7, CD10, CD13, CD19, CD20, CD33 and CD34 purchased from Becton Dickinson (Mountain View, San Jose, CA, USA). The proportion of positivity was expressed as a percentage with a cut-off of 20% over the corresponding isotypic control. Analysis was carried out first on an Epics Profile II (IL-Coulter, Hialeah, FL, USA) between 1990 and 1994, and then on an Epics XL flow cytometer (IL-Coulter) between 1995 and 2000. Intracytoplasmic and nuclear stains on cytospin preparations included cytoplasmic CD3 (cCD3) and CD22 (cCD22), cytoplasmic µ, SmIg and terminal deoxynucleotydil transferase (TdT), as previously described (Del Poeta et al, 1994). Immunofluorescence was evaluated by optical examination using a Zeiss microscope equipped with an epifluorescence set, a barrier filter set for blue fluorescence and a 100X phase contrast objective.

Cytogenetic analysis.  Karyotype analysis was performed on aspirated BM cells using standard Giemsa banding (G-banding) techniques to stain metaphase preparations obtained from unstimulated cultures, as previously described (Stasi et al, 1993). A successful karyotype analysis was obtained in 84 presentation samples. For the statistical analysis, patients were divided into two groups: favourable (diploid) and poor [hypodiploid, t(9;22), t(8;14), t(4;11) and other karyotypic abnormalities].

PGP and bcl-2 expression.  PGP expression was analysed in two-colour flow cytometric assays, using two primary pure mAbs: (1) C219 (Centocor Diagnostics, Malverne, PA, USA), IgG2a isotype and (2) JSB-1 (Bouty, Milan, Italy) that recognize different epitopes on the inner surface of the cytoplasmic membrane. Bcl-2 was detected by using an anti-bcl-2 124 FITC-conjugated mAb (Dako, Glostrup, Denmark). For the multiparameter analysis of PGP and bcl-2 combined with lymphoid antigens, mononuclear cells were first incubated with CD19 PE- or CD7 PE-conjugated mAbs for 30 min at 4°C. Subsequently, the cells were washed twice in phosphate-buffered saline (PBS), and were fixed and permeabilized in two steps: first in 3·5% paraformaldehyde/PBS and then in 50% cold acetone/PBS. All samples were incubated at 4°C for 30 min with either 10 µl of unconjugated C219 or 5 µl of pure JSB-1, or with 10 µl of bcl-2 FITC. For PGP (C219 and JSB-1), cells were further incubated at 4°C with a FITC-conjugated F(ab)2 fragment of goat anti-mouse Ig (Dako). Three doxorubicin (DX/DOX)-selected resistant cell lines (Lovo DX, 8226/Dox6 and 8226/Dox40) and their parental sensitive lines (Lovo 109, 8226/S) were used for this study as a reference for PGP expression (Dalton et al, 1989). Analysis was performed by flow cytometry as described earlier. Results were obtained by gating blast cells using SSC versus FSC, and examining C219, JSB-1 and bcl-2 expressions on either CD19- or CD7-positive blast cells (Fig 1). The results were expressed as ratios of mean fluorescence intensity (RMFI), obtained by dividing the mean fluorescence intensity (MFI) of the positively stained cells (C219, JSB-1 or bcl-2) by that of cells stained with their specific isotype control antibodies. Results are given as the mean ± standard deviation (SD). We attempted to set various cut-off points empirically for these biological variables, and the selected thresholds were sufficient to severely affect response and survival, leading to the accurate identification of patients with a poor prognosis. Moreover, we applied a discriminate function analysis based on the squared Mahalanobis distances of each case from its group centroids (C219, JSB-1 and bcl-2). The correct percentage (observed classification versus predicted classification) was more than 70% for all variables. The thresholds were set > 2 for C219, > 5 for JSB-1 and > 15 for bcl-2, using the lymphocytes from normal donors as negative controls and reference cell lines (Lovo DX for PGP and KG-1 for bcl-2) as positive controls.

Figure 1.

 Flow cytometric histograms of JSB-1 and bcl-2 expression on three example ALL samples. ALL cases were analysed by flow cytometry after exposure to anti-CD19 PE and anti-JSB-1 or anti-bcl-2 FITC mAbs, as described in the text. (A) A bivariate histogram of FSC/SSC with the ‘blast gate’ on leukaemic cells, IgG1-isotype-negative controls for JSB-1 (B) or bcl-2 (C) and CD19 mAbs from ‘blast gate’. Three examples of JSB-1 (D–F) or bcl-2 (G–I) mean fluorescence intensities (MFI) evaluated on the CD19+ blast cells segregated by the ‘ALL gate’. Three examples of JSB-1 (d–f) or bcl-2 (g–i) univariate histograms, conditioned on the ‘ALL gate’, showing the distribution of fluorescence, reported in figures as arithmetic means (M). JSB-1 and bcl-2 RMFI were calculated as the ratio of JSB-1 or bcl-2 mAbs mean divided by the negative control mean, as reported in the text.

PGP function.  PGP function was studied by measuring the retention of the fluorescent substrate Rhodamine 123 (Rh123), as previously described (Klimecki et al, 1994). In brief, blast cells were incubated with 150 ng/ml of Rh123 (Sigma, St Louis, MO, USA) for 15 min at 37°C in RPMI-1640 medium, washed and incubated with 10 ml of RPMI with 7% fetal bovine serum at 37°C with or without the presence of 25 µmol/l racemic verapamil (Sigma). Afterwards, the samples were taken at 0, 60, 90 and 120 min of incubation, washed at 4°C and incubated at 4°C for 30 min with an anti-CD19 PE or anti-CD7 PE mAb. After labelling, samples were maintained at 4°C and immediately analysed. Flow cytometric analysis was performed as described earlier. Rh123 MFI was measured as for FITC, and was determined by the rate of efflux with and without r-verapamil. The results were expressed as the ratio of the two means of fluorescence, with and without r-verapamil, at 90 min.

Statistical analysis.  The relationships of PGP and bcl-2 RMFI, used as continuous variables, to either biological or clinical predictors (age, karyotype, B/T-cell immunophenotype, CD34 or myeloid antigens) were assessed using the Mann–Whitney U-test. Correlations between PGP or bcl-2 RMFI, used as dichotomized variables, and the CR or relapse rates were based on the two-tailed Fisher's exact test. Correlations between C219 and JSB-1 RMFI values were performed by the Spearman's test. OS and DFS curves were plotted according to the method of Kaplan–Meier. OS was measured from the date of diagnosis to the date of either death or last follow-up. DFS was calculated from the date of CR until relapse. All the patients who underwent autologous or allogeneic BM transplantation were censored at the time of BM infusion. For the comparison of either DFS or OS patterns of two or more groups, the log-rank test was applied. The Cox proportional hazards regression model was used to assess the independent effect of covariables, treated as dichotomous, on OS and DFS. The variables for which univariate analysis had shown a significant association were studied.


PGP phenotype

The MFI staining for ALL patients was 2·3 ± 1·8 using C219 and 4·1 ± 3·2 using JSB-1 mAbs. The average MFI of the PGP staining (the average for both C219 and JSB-1 mAbs) for 8226/S and 8226/Dox controls was 1·8 ± 2 and 3·8 ± 4·1 respectively. Lymphocytes from 50 peripheral blood samples of normal donors were used as a negative control group: in all cases, PGP RMFI measured by flow cytometry was always lower than 2 for the C219 mAb and 5 for the JSB-1 mAb. With these cut-off points, 43·7% of the patients were C219 positive and 31·4% were JSB-1 positive. C219 RMFI and JSB-1 RMFI values were significantly and directly correlated (r = 0·39, P = 0·0008, Spearman's test). No significant association was observed between PGP levels and white blood cell (WBC) count, B/T-cell immunophenotype, CD34 or myeloid antigen expression. In contrast, a JSB-1 RMFI of higher than 5 was slightly correlated with t(9;22) [8/18; P = 0·04].

PGP function

Rh123 accumulation studies were performed both on lymphocytes of 50 peripheral blood samples from normal donors and in 62 ALL patients. Rh123 MFI ratios, obtained by dividing the Rh123 MFI observed with r-verapamil and Rh123 MFI observed without r-verapamil at 90 min, appeared to be significantly increased in either C219- or JSB-1-positive patients, when compared PGP-negative patients (P = 0·005 and P = 0·004 respectively) and normal donors, as shown in Table II.

Table II.   PGP function in lymphocytes from 50 normal donors, in 8226 reference cell lines and in 62 ALL patients.
 Rh 123 MFI Ratio ± SD*P-value
  • *

    Rh123 MFI ratio values were obtained by dividing Rh123 MFI with r-verapamil and Rh123 MFI without r-verapamil at 90 min.

  • †The P-value of the Rh123 MFI ratio of C219+ compared with C219 patients, and of JSB-1+ compared with JSB-1 patients is shown, determined using the t-test for independent samples.

Cell lines
 8226/S1·1 ± 1·2 
 8226/Dox61·5 ± 1·4 
 8226/Dox403·5 ± 2·9 
Normal donors (lymphocytes)1·05 ± 1·1 
Phenotype of ALL patients
 C219+ (26)1·9 ± 1·6 
 C219 (36)1·1 ± 0·90·005
 JSB-1+ (18)2·2 ± 2·3 
 JSB-1 (26)1·2 ± 1·10·004

Bcl-2 oncoprotein

The MFI of bcl-2 124 mAb staining in our ALL patients was generally lower (23·3 ± 19·2) than the KG-1 cell line mean (36·5 ± 11·4). Normal lymphocytes from 35 peripheral blood samples were used as negative control: in almost all cases the bcl-2 RMFI was lower than 15. With this cut-off point, 53·3% of the patients were bcl-2 positive. Significant associations were observed between lower bcl-2 RMFI levels and either an age > 45 years or an abnormal karyotype (P = 0·01 in both cases, Table III). A striking association was also found between a bcl-2 RMFI of less than 15 and either a B-mature immunophenotype (12/13; P = 0·001) or t(9;22) (13/21; P = 0·01).

Table III. 
(A) A comparison of the RMFI bcl-2 levels between samples from patients with an age either lower or higher than 45 years
 Age < 45 yearsAge > 45 yearsP-value*
RMFI bcl-2
 Number of samples5421 
(B) A comparison of the RMFI bcl-2 levels between samples from patients with either a normal or abnormal karyotype
 Normal karyotypeAbnormal karyotypeP-value*
  • *

    P-value as determined by Mann–Whitney U-test.

  • P25 indicates the 25th percentile; P75, the 75th percentile.

RMFI bcl-2
 Number of samples2339 

Clinical outcome

The most relevant clinical data in adult ALL patients are summarized in Table I. CR was achieved in 82·1% of the patients, with a median OS of 18·2 months and a median DFS of 18·7 months. No significant difference was found between PGP-positive and PGP-negative patients with regard to CR rate (77%vs 84% for C219; P = 0·56 and 90%vs 81% for JSB-1; P = 0·48). In contrast, bcl-2-positive patients showed a significantly higher CR rate (90%vs 66%; P = 0·01) than bcl-2-negative patients. Relapse was observed in 44·9% (49/109) of the evaluable patients and, interestingly, JSB-1-negative patients showed a longer time to relapse than JSB-1-positive patients (6·6 months vs 3·8 months; P = 0·04, Fig 2A). Both C219- and JSB-1-negative patients showed a significantly longer OS (23%vs 0% at 3 years, for C219, P = 0·011, Fig 2B, and 42%vs 0% at 1·5 years, for JSB-1, P = 0·004, Fig 2C) and DFS (29%vs 0% at 2 years, P = 0·006 for C219, Fig 2D, and 53%vs 0% at 8·5 months, P = 0·00006 for JSB-1, Fig 2E) than their positive counterparts. Moreover, we analysed 69 ALL patients not presenting with either t(9;22) or the B-mature immunophenotype. Within this subset, PGP negativity (JSB-1) maintained its significant favourable prognostic impact with regard to both OS (41%vs 0% at 1·5 years; P = 0·009, Fig 3A) and DFS (83%vs 0% at 6 months; P = 0·0005, Fig 3B), demonstrating its independence from other well-known risk factors in ALL. Importantly, within a subset of 62 ALL patients with either a normal (n = 31) or unknown (n = 31) karyotype, PGP (JSB-1)-negative patients also showed both a significantly longer OS and DFS than JSB-1-positive patients (63%vs 0% at 1·4 years, P = 0·018, Fig 3C, and 84%vs 0% at 6 months, P = 0·001, Fig 3D respectively).

Figure 2.

  (A) The time to relapse curve based on JSB-1 mean fluorescence intensity ratio (RMFI) values (n = 49 patients). A shorter time to relapse (3·8 months vs 6·6 months) characterized patients with a JSB-1RMFI above 5 (P = 0·04). (B) The OS and (D) DFS curves based on C219 and JSB-1 RMFI values (n = 115 patients). C219 RMFI values above 2 characterized patients with a shorter OS (0%vs 23% at 3 years, P = 0·011) and DFS (0%vs 29% at 2 years, P = 0·006). Equally, adult ALL patients with JSB-1 RMFI values above 5 had both a shorter OS [0%vs 42% at 1·5 years, P = 0·004 (C)] and DFS [0%vs 53% at 8·5 months, P = 0·00006 (E)]. P-values determined by the log-rank test are shown.

Figure 3.

 OS and DFS curves, based on JSB-1 mean fluorescence intensity ratio (RMFI) values, within a subset of patients not presenting t(9;22) or the B-mature immunophenotype (n = 69). Within this subset, PGP negativity (determined by JSB-1) maintained its significant favourable prognostic impact with regard to both OS [41%vs 0% at 1·5 years, P = 0·009 (A)] and DFS [83%vs 0% at 6 months, P = 0·0005 (B)]. Moreover, in ALL patients with a normal or unknown karyotype (n = 62 patients), we performed OS and DFS curves, basing on JSB-1 RMFI values. Within this subset of ALL patients with a normal (n = 31) or unknown (n = 31) karyotype, PGP (determined by JSB-1)-negative patients showed both a significantly longer OS [63%vs 0% at 1·4 years, P = 0·018, (C)] and DFS [84%vs 0% at 6 months, P = 0·001 (D)]. P-values determined by the log-rank test are shown.

No significant correlation was found between bcl-2 levels and either OS or DFS. Moreover, in the univariate analysis of DFS, neither age (P = 0·4), B/T-cell immunophenotype (P = 0·2), nor peripheral blast cell count (P = 0·5) achieved statistical significance. Therefore, in the multivariate modelling, only JSB-1 (P = 0·008) and cytogenetics (P = 0·02) were found to be independent prognostic factors with regard to DFS, as shown in Table IV.

Table IV.   The multivariate analysis of DFS in adult ALL patients.
VariablesHazard ratioP-value*95% confidence interval
  • *

    The P-value was determined using Cox proportional hazards regression model for DFS.



The expression of PGP and bcl-2 was studied in newly diagnosed adult ALL patients and was related to different risk indicators (WBC count, immunophenotype, age and karyotype).

PGP phenotype was analysed by flow cytometry using two mAbs (C219 and JSB-1) and MFIs. In ALL cases, C219 positivity was found in 43·7% of patients and JSB-1 positivity in 31·4% of patients. Because both of these antibodies react with intracellular epitopes, permeabilization is required. As Goasguen et al (1993) and Savignano et al (1993), we found that paraformaldehyde and acetone fixation was optimal for permeabilization of the cell membrane, giving access to the internal PGP epitope recognized by JSB-1 without causing modification of the antigen. Den Boer et al (1997) compared 13 fixation methods using six antibodies and both immunocytochemistry and flow cytometry techniques for PGP, multidrug-resistance protein (MRP) and lung-resistance protein (LRP) in childhood ALL. The optimal fixation for PGP (C219, MRK16), MRP (MRPr1) and LRP (LRP56) was a mixture of 2% (v/v) formaldehyde solution and acetone incubated for only 10 s at room temperature (formaldehyde and acetone).

In our study, the C219 RMFI was strongly related to the JSB-1 RMFI (P = 0·0008, Spearman's test). The different percentages of positivity of these two mAbs (C219 and JSB-1) may be explained by the fact that C219 recognizes both MDR-1 and MDR-3 products (Schinkel et al, 1991), whereas JSB-1 is specific for MDR-1 only (Scheper et al, 1988).

The PGP expression did not significantly differ between risk groups of patients identified by immunophenotype, WBC and age. On the other hand, only a slight correlation between higher PGP levels and t(9;22) was found when using the JSB-1 mAb (P = 0·04).

As expected, an increased Rh123 MFI ratio showed a significant correlation with PGP positivity (Table II), although a proportion of ALL patients defined as PGP positive by either C219 or JSB-1 mAbs, presented low Rh123 MFI ratios. Ludescher et al (1993) suggested that in ALL, PGP might be expressed, but non-functional in some patients. Wuchter et al (2000a) determined PGP function by the Rh123 efflux test in 102 children with de novo ALL, and found a discrete correlation between PGP phenotype, a lower functional PGP activity than that observed in AML and no clinical significance.

Several studies based on the evaluation of MDR-1 in ALL have shown a correlation between MDR and a resistance to treatment. To elucidate the role of the PGP efflux pump, many investigators have focused their studies on the PGP cytoplasmic membrane antigen. An analysis of PGP expression by JSB-1 in 59 ALL patients (36 children, 23 adults) found a lower CR rate in MDR+ adult patients only, and a higher relapse rate among MDR+ patients was observed in both adults and children (Goasguen et al, 1993). A multicentric study of 203 adult ALL patients, performed using flow cytometry, demonstrated that MDR1 protein expression was an independent predictor of CR achievement (Tafuri et al, 2002). Significantly higher levels of protein expression have been found in recurrent relapses of ALL (Beck et al, 1996). However, in 95 de novo adult ALL patients, Damiani et al (2002) observed that the CR rates were similar in MDR-positive and -negative patients, but that relapses within 6 months were more frequent in PGP+ patients, and also that the DFS was shorter in PGP+ compared with PGP patients (P = 0·01). Our study reports upon one of the largest series of de novo adult ALL patients with a long follow-up, examined on the basis of PGP-associated membrane protein. A short median time to relapse (5·6 months, Table I), a poor OS (20% at 11 years, Table I) and a short DFS (25% at 11 years, Table I) were observed in ALL patients. In our opinion, the reason for these results are threefold: (a) a very high number of Philadephia-chromosome-positive (Ph+) ALL (35%) and B-mature ALL patients (15%) were treated with conventional schemes, (b) the high median age (> 40 years) of ALL patients, and finally (c) a WBC count higher than 30 × 109/l in approximately 40% of patients. These concurrent and adverse prognostic factors did not allow us to achieve the leukaemia-free survival rates of 30–40% described in literature (Hoelzer et al, 1993; Durrant et al, 1997; Thiebaut et al, 2000). However, the GIMEMA ALL 0288 trial (Annino et al, 2002), which was also a large co-operative study, presenting a median age of 27·5 years and excluding B-mature ALL, reported a median time to relapse of only 10 months and an OS of 27% at 9 years. Intensified consolidation, particularly with high-dose methotrexate and high-dose cytarabine, could be one reason for the improved outcome in recent studies (Kantarjian et al, 2000; Bassan et al, 2001). However in adult ALL, it seems that both the intensification of chemotherapy and increased use of stem cell transplantation mean that the opportunity for the further improvement of patient prognosis is limited. This is particularly true for the high proportion of elderly ALL patients who are not included in many trials. New options for treatment include molecular targeting with kinase inhibitors, antibody therapy, non-myeloablative transplantation and the application of donor leucocyte infusions (Hoelzer & Gökbuget, 2000).

An important result of our study was the significantly shorter time to relapse (P = 0·04, Fig 2A) observed in JSB-1+ ALL patients. Moreover, PGP-negative patients had a significantly longer OS and longer DFS than PGP-positive patients, as shown in Fig 2B–E. Thus, we can postulate a primary role of the MDR phenotype in ALL, not only in the prediction of early relapse, but also in a poorer outcome. The independent prognostic significance of PGP with regard to OS and DFS was statistically confirmed within two subsets: patients not presenting either t(9;22) or B-mature phenotype (Fig 3A–B), and those with a normal or unknown karyotype (Fig 3C–D).

Defects in the mitochondrial-mediated pathways of apoptosis can also induce multidrug resistance and may offer prognostic value in ALL. The over-expression of bcl-2 is an important cause of multidrug resistance. In AML, high levels of bcl-2 have been correlated with lower CR rates and a lower OS (Campos et al, 1993). However, other studies have indicated that increased levels of bcl-2 are neither prognostic nor associated with an improved survival. In patients with breast cancer, increased bcl-2 levels detected by immunostaining have been correlated with an improved DFS (Charpin et al, 1998), and in childhood ALL increased bcl-2 expression has been associated with an improved event-free survival (Coustan-Smith et al, 1996). Moreover, in childhood ALL, Wuchter et al (2000b) found that good responders to initial prednisone therapy had significantly higher bcl-2 expression levels than poor responders (P = 0·002). These results may reflect the importance of post-translational modification of bcl-2. Several studies have suggested that phosphorylation of bcl-2 affects its function, although there have been conflicting reports as to whether this enhances or reduces its ability to suppress apoptosis (Haldar et al, 1995). It is also possible that mutations may alter the function of bcl-2 by affecting protein stability. In agreement with these observations, our study provides novel flow cytometry data on the expression of the bcl-2 protein in adult ALL, demonstrating that bcl-2+ patients had a significantly higher CR rate, but not significantly longer OS and DFS. However, in the multivariate analysis of DFS, bcl-2 determination had a trend towards statistical significance (P = 0·06). In a larger series of adult ALL patients, the prognostic value of the bcl-2 protein for DFS may be identifiable. It could be proposed that the resistance mechanisms related to PGP, rather than the apoptotic defects related to bcl-2, are more closely related to both the chemotherapeutic failure and the worse outcome of adult ALL.

In conclusion, using multivariate analysis, both PGP (P = 0·008) and cytogenetics (P = 0·02) were confirmed as independent prognostic factors for the DFS of ALL patients. Therefore, our study suggests that the assessment of multidrug resistance mechanisms, based both upon both PGP and bcl-2 and performed by flow cytometry, may be an important predictor of treatment outcome in adult ALL.


We would thank the members of our Department of Haematology clinical staff for their support to our ALL clinical research program.