Professor Fenaux Service des Maladies du Sang, CHU, 1 place de Verdun, 59037 Lille, France.
Intensive chemotherapy produces a lower complete remission (CR) rate in the myelodysplastic syndromes (MDS) than in de novo acute myeloid leukaemia (AML), possibly due in part to a higher incidence of P glycoprotein (PGP) expression in MDS blast cells. We designed a randomized trial of intensive chemotherapy with or without quinine, an agent capable of reverting the multidrug resistance (mdr) phenotype, in patients aged 65 years with high-risk MDS. Patients were randomized to receive mitoxantrone 12 mg/m2/d days 2–5 + AraC 1 g/m2/12 h days 1–5, with (Q+) or without (Q−) quinine (30 mg/kg/d). 131 patients were included. PGP expression analysis was successful in 91 patients. In the 42 PGP-positive cases, 13/25 (52%) patients in the Q+ group achieved CR, compared to 3/17 (18%) patients in the Q− group (P = 0.02) and median Kaplan-Meier survival was 13 months in the Q+ group, and 8 months in the Q− group (P = 0.01). No life-threatening toxicity was observed with quinine. In conclusion, the results of this randomized study show that quinine increases the CR rate and survival in PGP-positive MDS cases treated with intensive chemotherapy.
The decreased intracellular accumulation of a variety of cytotoxic agents, characteristic of the mdr phenotype, can be reversed in vitro by a number of non-cytotoxic agents but the in vivo use of most of these agents is precluded by serum protein binding or clinical toxicity ( Bennis et al, 1995 ; Dorr & Liddil, 1991; Genne et al, 1994 ; Malayeri et al, 1996 ; Solary et al, 1990 ; Sonneveld, 1996; Wigler & Patterson, 1994; Pajeva et al, 1996 ). Intravenous infusion of conventional doses of quinine achieves a sufficient serum concentration to reverse the anthracycline resistance in mdr models, and can be safely administered in combination to mitoxantrone and AraC ( Solary et al, 1991a , b, 1992). In the preliminary results of a recent randomized trial, the French Groupe Ouest Est des Leucémies Aiguës Myéloïdes (Goelams) showed that the addition of quinine to mitoxantrone–AraC chemotherapy could improve the CR rates in AML, particularly in AML post-MDS ( Solary et al, 1996 ). Differences were not significant, including for AML post-MDS, but the number of patients in this subgroup was relatively small, and PGP expression had been studied in a limited number of centres. Thus the Goelams group and the Groupe Français des Myélodysplasies (French MDS group: GFM) performed a similar randomized study of mitoxantrone-AraC chemotherapy and quinine in MDS and AML post-MDS. Randomization was stopped after the second interim analysis, as the CR rate in cases that expressed PGP was significantly superior with the addition of quinine.
PATIENTS AND METHODS
Between March 1992 and May 1996, the GFM and the Goelams conducted a randomized trial of mitoxantrone–AraC chemotherapy with or without quinine in MDS and AML post-MDS (MAQ trial). This trial had been approved by the ethics committee of Dijon (France) according to French law. Inclusion criteria were (1) MDS, according to FAB criteria, (2) age 65 years, (3) marrow blasts 5%, i.e. refractory anaemia with excess blasts (RAEB), RAEB in transformation (RAEB-T) chronic myelomonocytic leukaemia (CMML) or MDS having progressed to AML (AML post-MDS), (4) patient diagnosed in a centre where PGP expression was assessable or adequate material sent to a referral centre, (5) written informed consent of the patient. Randomization was made by a telephone centralized procedure. It was not stratified on result of PGP expression, as this result was not immediately available in many centres.
Patients allocated to chemotherapy alone (control group) received an induction course of mitoxantrone (MXN) 12 mg/m2/d, days 2–5 and AraC 1 g/m2/12 h, days 1–5. Patients allocated to the quinine group received the same MXN–AraC regimen with quinine formiate (Quinimax; Labaz Laboratories, Paris, France) at 30 mg/kg/d, started 24 h before the first dose of MXN and administered by continuous i.v. infusion until 24 h after the end of the last MXN infusion. Toxicity was assessed according to the W.H.O. grading system.
Patients < 55 years with no HLA-identical sibling and who achieved CR were scheduled to receive autologous bone marrow or blood stem cell transplantation. Peripheral stem cell harvest was performed after a consolidation chemotherapy (CT) course consisting of MXN 12 mg/m2 days 1–2, AraC 500 mg/m2/12 h days 1–4 followed by G-CSF 10 μg/kg/24 h. Older patients received three to six consolidation courses of MXN 12 mg/m2 day 1 and AraC 60 mg/m2/12 h days 1–5.
In AML post-MDS, PGP expression was determined by flow cytometry using MRK16 antibody (Immunotech, France), a monoclonal antibody which recognizes an extracellular determinant of P-glycoprotein, in a previously described indirect immunofluorescent assay ( Solary et al, 1991b ). Briefly, cells were fixed in 1% paraformaldehyde for 30 min at 4°C and incubated in human AB serum before a 30 min incubation with MRK16 antibody (2 μg/ml), then with fluorescein-conjugated F(ab′)2 fragments of goat antimouse IgG (37.5 μg/ml; Silenius laboratories). The percentage of positive cells was determined by flow cytometry using a Becton Dickinson FACScan flow cytometer. A non-relevant IgG2a (Coulter, Raritan, N.J.) was used with the same goat antimouse second-step reagent to calibrate the assay on the FACScan and determine the positivity for individual cells. P-glycoprotein-positive K562/ADM cells were used as positive control. MRK16 analysis was performed by six different centres, which belonged to the French Drug Resistance Intergroup, using an agreed system for reporting results and development of quality control ( Marie et al, 1997 ). Results were expressed as a ratio of the arithmetic mean of fluorescence for antibody/arithmetic mean of fluorescence for control. PGP expression was considered positive if the ratio was > 1.5 ( Marie et al, 1997 ).
In non-transformed MDS patients, PGP expression was assessed by immunocytochemistry on marrow slides with JSB1 MoAb (Monosan, Uden, The Netherlands) using the alkaline phosphatase–antialkaline phosphatase (APAAP) technique or the avidin–biotin–peroxidase technique, as previously described ( Lepelley et al, 1994 ). A sample was considered positive when at least 5% of blasts were stained with JSB1 in the absence of control staining. In our experience, results of immunocytochemistry with JSB1 MoAb and flow cytometry results with MRK16 MoAb are highly correlated in MDS and AML ( Poulain et al, 1998 ).
The achievement of complete remission (CR) in cases with PGP expression, was the major endpoint. CR criteria included: (i) marrow blasts < 5%; (ii) neutrophils > 1.5 × 109/l, platelets > 100 × 109/l, haemoglobin > 10 g/dl without transfusion requirement; (iii) normalization of the karyotype if initially abnormal (at least 10 mitoses examined); (iiii) disappearance of major myelodysplastic features in the bone marrow (with the exception of mild dysplasia, as seen after chemotherapy). Partial remission (PR) criteria included: < 5% marrow blasts and persistence of cytopenia(s) but with an increase in neutrophils of at least 1 × 109/l, platelets of at least 50 × 109/l, of haemoglobin of at least 2 g/dl, or disappearance of cytopenias but persistence of either major myelodysplastic features or cytogenetic abnormalities. Death in aplasia was defined by death during the period of aplasia following chemotherapy administration. Failure corresponded to all other situations.
The CR duration and survival in with PGP-positive cases; the CR rate, CR duration and survival of the overall population and in PGP-negative patients were considered as secondary endpoints.
Patients were included from 22 French centres that could perform PGP expression analysis or sent material to a reference centre. A first interim analysis was performed after inclusion of 90 patients, at the reference date of 1 May 1994. The trial was stopped after the second interim analysis, performed after the inclusion of 131 patients, at the reference date of 1 May 1996. A third interim analysis, whose results are presented here, was performed at the reference date of 30 May 1997. Analysis was made on an intention-to-treat basis.
Prognostic factors were assessed using the appropriate regression models, either logistic regression or Cox model. These factors, derived from previous studies ( Fenaux et al, 1991 ; Morel et al, 1993 ) were age, sex, FAB classification, WBC count, haemoglobin level, platelet count, absolute number of circulating blasts, percentage of bone marrow blasts, cytogenetic abnormalities and time from diagnosis to treatment.
Termination of the trial
The second interim analysis, performed at the reference date of 1 May 1996, found a significantly higher CR rate, among PGP-positive patients who had received quinine, and a decision to stop patient accrual was made. At that time, 131 patients had been randomized. 91 of them (70%) had interpretable PGP expression analysis and 42 were PGP positive.
Initial characteristics of the patients
1 Table I summarizes the initial characteristics of the 131 patients. No significant differences in pretreatment characteristics were found between the two groups.
Table 1. Table I. Initial characteristics of the 131 patients. ns: not significant.
PGP expression analysis was successful in 91 patients. 42 patients (46%) had positive PGP expression and 49 (54%) were PGP negative. In the remaining 40 patients (30%) PGP expression could not be evaluated, mainly due to the absence or poor quality of cellular samples, or rarely to laboratory technical problems. There was a trend for higher incidence of PGP expression in males (35/68, 52%) than in females (7/24, 29%) but the difference was not significant (P = 0.07).
Table 4. Table IV. Response to treatment and evolution. ns: not significant.
Of the 131 patients, 57 (44%) achieved CR, 22 (17%) PR, 30 (23%) had resistant disease, and 22 (16%) died during aplasia. Of the 91 patients in whom PGP expression was successfully assessed, 37 (41%) achieved CR, 20 (22%) PR, 19 (21%) had resistant disease, and 15 (16%) died during aplasia. The CR rate was 38% in PGP-positive cases and 43% in PGP-negative cases (difference not significant).
Of the 62 patients allocated to the quinine group, 29 (47%) achieved CR, as compared to 28/69 (41%) patients treated with chemotherapy alone (difference not significant). Similarly, there was no significant difference in the incidence of death during aplasia between the two groups: 13% in the quinine arm versus 20% in the control arm.
In PGP-positive cases, 13/25 (52%) patients who received quinine achieved CR, as compared to 3/17 (18%) patients treated with chemotherapy alone (P = 0.02, Mantel-Haenszel test). On the other hand, in PGP-negative cases, the CR rate was 35% and 49%, respectively, in patients who received quinine or chemotherapy alone (difference not significant).
CR duration and survival
Median follow-up of the 131 patients was 9 months. Median Kaplan-Meier estimate of overall survival was 12 months. Of the 57 patients who achieved CR, 34 received consolidation CT, 20 were autografted (using bone marrow or peripheral stem cells harvested after CR achievement), and three were allografted in first CR. Allografted patients were censored at the time of transplantation. 36 patients relapsed after 2–37 months and 21 were still in CR after 7–59 months, with a median CR duration of 14 months. Median Kaplan-Meier estimate of survival was 13 months in the 62 patients allocated to the quinine group and 11 months in the 69 patients allocated to the control group (difference not significant, Fig 1).
In the 42 PGP-positive patients, the median Kaplan-Meier estimate of survival was 13 months in patients allocated to the quinine group, and 8 months in patients treated with chemotherapy alone (P = 0.01, Fig 2). In PGP-negative patients, median Kaplan-Meier estimate of survival was 14 months in patients allocated to the quinine group, and 14 months in patients treated with chemotherapy alone (difference not significant, Fig 3).
Toxicity of the regimen
Quinine toxicity was observed in 17/62 quinine-treated patients (27%) and included tinnitus (13/62, 21%), vertigo (8/62, 13%), bradycardia (1/62: 2%), QT interval increase on ECG (4/62, 6%), and mild hearing loss (6/62, 10%).
In 15 patients quinine-induced side-effects decreased or disappeared after a 20% quinine dose decrease. In two patients quinine infusion had to be stopped after 2 d due to excessive QT increase (one case) or hearing loss (one case).
Haematological toxicity is detailed in 5 Table V. In patients who achieved CR the mean duration of leucopenia, neutropenia and thrombocytopenia were longer in patients allocated to the quinine arm. However, this difference was not significant. There was a significantly prolonged duration of hospitalization in patients treated with quinine.
Table 5. Table V. Haematological toxicity according to induction treatment in patients who achieved CR.
Non-haematological toxicity is detailed in 6 Table VI. There was no significant difference in the incidence of > W.H.O. grade 2 diarrhoea, vomiting, nausea, liver and heart toxicity between the two arms. Mucositis was significantly more frequently observed in the quinine group ( 6 Table VI).
Table 6. Table VI. Non-haematological toxicity (> W.H.O. Grade 2) according to induction treatment. ns: not significant.
We found no significant difference in CR rate, CR duration, and surival between chemotherapy alone and chemotherapy combined with quinine in the overall patient population. However, in PGP-positive cases a significantly higher CR rate, more prolonged CR duration, and better survival were observed in patients treated with quinine when compared to patients treated with chemotherapy alone. A possible explanation could be an effect of quinine on the metabolism of mitoxantrone and/or AraC. Indeed, several drugs currently under investigation as mdr-reversing agents, including PSC833, can modify the metabolism of antineoplastic agents, including anthracyclines and epipodophyllotoxins ( Gonzalez et al, 1995 ; Sonneveld et al, 1996 ). This effect can result pharmacologically in an increase in drug concentration and possibly in higher efficacy (and also toxicity). However, in the present study, quinine had no effect on the response rate, CR duration and survival in PGP-negative MDS. Furthermore, in previous studies of the GOELAMS group, serum levels of quinine achieved by continuous infusion of this drug were sufficient to circumvent the mdr phenotype of a mdr-positive cell line ( Solary et al, 1991b , 1992, 1996). These findings strongly suggest a specific effect of quinine on PGP-mediated drug resistance rather than a simple effect on drug metabolism.
To our knowledge, this study is the first to demonstrate some benefit of the addition of a drug-reversing agent to antineoplastic drugs in a randomized trial. Quinine, like verapamil ( Dalton et al, 1995 ) and cyclosporine A ( Weber et al, 1995 ), belongs to the first generation of resistance modifiers. Several phase I/II trials with verapamil and cyclosporine A agents in relapsing AML ( List et al, 1993 ), multiple myeloma ( Weber et al, 1995 ) and non-Hodgkin's lymphoma ( Miller et al, 1991 ; Malayeri et al, 1996 ) have suggested that these drugs could yield some benefit in combination with anthracyclines, vinca alkaloids or epipodophyllotoxins. However, those studies were not randomized.
The favourable effect of quinine on PGP-mediated drug resistance obtained in the present study could be due to a better efficacy of quinine, as compared to many of the other potential reversing agents tested so far. Apart from our group, quinine has not to our knowledge been assessed clinically as a reversing agent by other groups. On the other hand, quinine – and possibly other reversing agents – could have a greater reversing effect on PGP-positive blasts from MDS patients than on PGP-positive tumour cells from other neoplasms. In the case of AML, preliminary results from the GOELAMS group already suggested a greater effect of quinine on the drug sensitivity of blasts from AML occurring after MDS and myeloproliferative disorders than on blasts from de novo AML cases ( Solary et al, 1996 ).
Second-generation resistance modifiers such as PSC833 ( Gonzalez et al, 1995 ), dexverapamil ( Wilson et al, 1995 ) or cinchonine ( Genne et al, 1992 ) have been designed to increase specificity for PGP expression and to decrease toxicity. In vitro studies have shown that these drugs allowed sufficient modulation of P-glycoprotein under serum conditions at concentrations achievable in vivo, and clinical studies with these drugs are underway ( List et al, 1996 ). Because our results suggest that reversing agents may be clinically useful in MDS, it will be important to test these drugs in this group of disorders and to compare their clinical usefulness with that of quinine.
Side-effects of quinine mainly included tinnitus and vertigo, which generally disappeared with 20% dose reduction. No life-threatening cardiac toxicity was observed. There was a trend for more prolonged duration of neutropenia and a significantly prolonged duration of hospitalization in patients treated with quinine. In addition, quinine induced a significant increase in the incidence of mucositis.
In addition to PGP, other molecules, including multidrug resistance-associated protein (MRP) ( Beck et al, 1996 ) and lung resistance protein (LRP) ( List et al, 1996 ), interfere with the transport of cytotoxic drugs. LRP expression, in particular, is observed in about 40% of MDS and is only partially correlated with PGP expression ( List et al, 1996 ; Lepelley et al, 1998 ). Little is known about MRP expression in MDS. Finally, in addition to alteration of drug transport, drug resistance may arise from alterations at any other step in the cell-killing pathway including drug metabolism ( Holmes et al, 1990 ), drug target and DNA repair mechanisms ( Gieseler et al, 1996 ), and in the ability of cells to recognize a toxic insult and engage apoptosis. Some of these mechanisms have prognostic implications in AML and/or MDS treated with intensive chemotherapy ( Campos et al, 1993 ; Wattel et al, 1994 ) and may provide important targets in the modulation of drug resistance.
This work was supported by the Programme Hospitalier de Recherche Clinique (Centre Hospitalier Universitaire of Lille).