INTRODUCTION

The hallmark of the multidrug resistance phenotype is cross-resistance to multiple compounds that are unrelated in structure, cellular target, and mode of action (1). Overexpression of the protein product of the multidrug resistance gene (MDR-1), P-glycoprotein, is a major, as well as the best understood, mechanism of multidrug resistance. It is a large, integral membrane protein that seems to function as an adenosine triphosphate–dependent, selective drug efflux pump (2) and confers resistance to a variety of drugs from different mechanistic and structural classes (3).

As patients with acute erythroleukemia, particularly the M6b and M6c subtypes, characteristically demonstrate an unusually poor response to chemotherapeutic agents (4, 5, 6), overexpression of the MDR-1 product was postulated to play an important role. We therefore performed immunohistochemical studies for P-glycoprotein on 48 archival cases of acute erythroleukemia and correlated these findings with clinical presentation, cytogenetic analyses, chemotherapeutic regimens, follow-up, and survival.

MATERIALS AND METHODS

Case Material Procurement

Forty-eight archival cases of acute erythroleukemia, diagnosed between 1984 and 1997, were available for review and performance of additional immunohistochemical stains; 14 from the files of the University of Cincinnati Medical Center, 22 from Loyola University Medical Center, 7 from Miami Valley Hospital, 4 from the National Naval Medical Center, and 1 from Scottish Rite Children’s Hospital in Atlanta.

Case Classification

The cases were divided into three groups, based on a 500-cell count performed on the bone marrow aspirate smears, as previously described (6). All cases had normal serum vitamin B12 and folate levels and characteristically displayed more than 50% erythroid component. The cases were subdivided into groups: M6a (23 cases corresponding to the traditional FAB-M6 category, demonstrating at least 30% myeloblasts of the nonerythrocytic component), M6b (15 cases of pure erythroleukemia, with >30% pronormoblasts of the erythrocytic series), and M6c (10 cases with >30% myeloblasts of the nonerythrocytic elements and >30% pronormoblasts of the erythrocytic component). Morphologic interpretation was confirmed with immunohistochemical stains for myeloperoxidase and hemoglobin, and archival flow cytometric analyses when available.

Case Analysis

Review of pertinent clinical information, including medical history, toxin exposure (7, 8), alcohol use, cytogenetic analyses, prior and subsequent use of chemotherapy, and survival from the time of diagnosis of acute leukemia, was performed for all cases through chart review and contact with the Tumor Registry of each participating institution, as previously described (6).

Cytogenetic analyses were divided into favorable [t(8;21); inv 16; t(16;16); +14], unfavorable [−5/5q−; −7/7q−; inv 3; 11q abnormalities; 17p abnormalities or i(17q); del 20q; +13; t(9;22); dmins; multiple cytogenetic aberrations (>2 abnormalities)], and intermediate [normal karyotype and all other abnormalities], as previously described (9).

Immunohistochemistry

Immunohistochemical stain for P-glycoprotein (NCL-pGLYp, Ventana 1:75, Novacastra, Newcastle upon Tyne, UK) was performed according to the method of Hsu et al. (10), after antigen retrieval as per Kawai et al. (11). All slides were reviewed for quality, and positivity (strong versus weak) or negativity of the leukemic populations was determined. In keeping with the literature, cases with at least 10% of cells reacting with the immunostain were deemed “positive.” The intensity of the stain was concluded to be either “strong” or “weak.”

Statistics

The blinded results were tabulated according to subtype, and univariant statistical analyses were performed (confidence intervals, χ2 analysis, Komolgorov-Smirnov analysis, linear regression) using the StatMost statistical program (DataMost Corp., Salt Lake City, UT). Multivariant analysis was not performed because of the small numbers of cases in each group.

RESULTS

Clinical Data

Demographic data, significant medical history, cytogenetic analyses, and clinical presentation of these patients have been reported (6). The pertinent clinical findings for each patient are summarized in Table 1. Significant is that acute erythroleukemia has been shown to evolve frequently from a prior myelodysplastic syndrome, suggesting that it is frequently a therapy-induced or myelodysplastic syndrome–derived acute leukemia (6, 12). In addition, approximately half of all patients with acute erythroleukemia of all subtypes have a history of toxin and/or alcohol exposure (6). Analysis of these poor prognostic indicators and their relationship to the various subtypes has been reported (6, 12).

TABLE 1 Relevant Clinical Data
Full size table

A favorable karyotype was not present in any of the patients in the present study. An intermediate karyotype was present in 10 of 16 M6a cases (62.5%), 1 of 11 M6b cases (9%), and 2 of 4 M6c cases (50%). An unfavorable karyotype was identified in 6 of 16 M6a cases (37.5%), 10 of 11 M6b cases (91%), and 2 of 4 M6c cases (50%).

Thirty-eight patient records contained treatment histories. Of these, 23 patients (60.5%) had received the standard myeloid protocol of high dose ara-C followed by daunorubicin (12 M6a, 9 M6b, 2 M6c). Other therapies included symptomatic relief only (5 M6a, 1 M6b, 2 M6c), bone marrow transplantation (1 M6a, 1 M6b, 1 M6c), high-dose ara-C only (1 M6b), erythropoietin (1 M6a), doxorubicin (1 M6b), and an unknown chemotherapeutic regimen (1 M6b). Remission in the M6a and M6c groups was achieved in all patients who received ara-C/daunorubicin or a bone marrow transplant (100%). In the M6b group, remission was obtained in only two patients who were treated with ara-C/daunorubicin (Patients 9 and 15).

Overall survival for each subtype was as follows: M6a, 31.4 ± 32 mo (median survival, 24 mo); M6b, 3.15 ± 4.2 mo (median survival, 1.75 mo); M6c, 10.5 ± 12.7 mo (median survival, 7 mo). Patient prognosis in the M6a group, divided according to P-glycoprotein status (Table 2), revealed mean survival of 11.2 ± 9.8 mo for the strongly reactive stain (median survival, 8.5 mo), 17.4 ± 10.2 mo for the weakly reactive stain (median survival, 19 mo), and 47.5 ± 38.7 mo for the negative stain (median survival, 35 mo). Statistical analysis for survival was performed for strong versus negative stain (P =.01), weak versus negative stain (P =.026), and strong versus weak stain (P =.2). The patients with M6b showed a mean survival of 1.6 ± 1.03 mo for the strongly reactive stain (median survival, 1.5 mo) and 9.3 ± 7.4 mo (median survival, 9.25 mo) for the weak immunostain (P =.08). Mean survival for the patients with M6c was 4.5 ± 3.5 mo for the strong (median survival, 4.5 mo) and 12.5 ± 14.2 mo (median survival, 9.25 mo) for the weak stain (P =.12).

TABLE 2 Survival Data for Each Subtype, Divided According to Treatment Status and P–Glycoprotein Expression
Full size table

Bone Marrow Findings

The histologic features and staining patterns for the bone marrow aspirates and biopsies of these cases have been reported (6, 12).

Immunohistochemical stain for P-glycoprotein revealed that 12 of 23 samples of patients with M6a were negative for protein expression (52.3%) (Fig. 1), whereas 6 patient samples stained weakly (26%) and 5 demonstrated a strong positivity (21.7%). The samples of patients with M6b all stained positive for P-glycoprotein (Fig. 2); 13 of 15 stained strongly (86.7%) and 2 stained weakly (13.3%) positive. Findings in the M6c subtype revealed 4 of 10 (40%) with a strong stain and 6 with a weak immunostain (60%) (Fig. 3). Positivity was predominantly within the less mature, blastic-appearing cells.

FIGURE 1
figure 1

Representative photomicrograph of acute erythroleukemia, M6a (Wright-Giemsa stained aspirate smear, 1000 × oil immersion).

Full size image
FIGURE 2
figure 2

Representative photomicrograph of acute erythroleukemia, M6b (Wright-Giemsa stained aspirate smear, 1000 × oil immersion).

Full size image
FIGURE 3
figure 3

Representative photomicrograph of acute erythroleukemia, M6c (Wright-Giemsa stained aspirate smear, 1000 × oil immersion).

Full size image

As a previous report demonstrated a negative additive effect of P-glycoprotein expression and unfavorable cytogenetic abnormalities (9), acute erythroleukemia patients with cytogenetic analyses and complete clinical data (survival, chemotherapy, and remission) were categorized according to P-glycoprotein expression and cytogenetic categories, as shown in Table 3 and Figure 4. Remission was achieved in all patients with negative and weakly positive immunohistochemical stain for P-glycoprotein. However, cases with a strongly positive stain demonstrated a decrease in the remission rate, depending on cytogenetic analysis: intermediate cytogenetics, one of two cases; unfavorable cytogenetics, zero of five cases. Statistical analyses on survival were performed for intermediate cytogenetics/negative P-glycoprotein stain and intermediate cytogenetics/weak P-glycoprotein stain (P =.02), as well as intermediate cytogenetics/negative P-glycoprotein stain and unfavorable cytogenetics/strongly positive stain (P =.002).

TABLE 3 Remission Rate and Mean Survival Differences of Treated Patients Based on P-Glycoprotein Expression and Cytogenetics
Full size table
FIGURE 4
figure 4

Schematic diagram of mean survival, based on P-glycoprotein expression and cytogenetics. Light grey, intermediate cytogenetics; dark grey, unfavorable cytogenetics.

Full size image

DISCUSSION

Acute erythroleukemia is a relatively rare disorder of a multilineal nature. By traditional FAB criteria, the myeloblasts were believed to be the malignant component. However, recent studies (4, 5, 6) have proved that the pronormoblasts play a key role in the poor response to chemotherapy and short survival characteristic of this disorder. The dismal outlook for these patients, particularly the M6b and M6c subtypes, may also be partially explained by frequent p53 mutations within diagnostic bone marrow at initial diagnosis (12).

Besides its role in unregulated proliferation and clonal expansion, mutated p53 has been shown specifically to stimulate the MDR-1 promoter, whereas wild-type p53 represses it (13, 14). P53 mutation does not lead to generalized drug resistance but rather causes selective resistance to P-glycoprotein substrates and increases sensitivity of the tumor cells to other drugs, such as methotrexate (13). An extensive study demonstrating this decreased growth inhibition by chemotherapeutic agents (drug resistance) in the presence of p53 mutations was recently reported (15).

P-glycoprotein is a transmembrane protein that functions as an energy-dependent drug efflux pump for selective resistance to anthracyclines (daunorubicin and doxorubicin), vinca alkaloids, podophyllotoxins (etoposide), antimicrotubule agents (paclitaxel and colchicine), and actinomycin D (1, 13, 16, 17). This multidrug resistance may be reversed or overcome by such agents as calcium-channel blockers, analogs of anthracyclines and vinca alkaloids, steroids and hormonal analogs, cyclosporine analogs, and miscellaneous hydrophobic cationic compounds (3, 18, 19, 20). MDR-1 gene expression in acute leukemias has been documented extensively (15, 16, 21, 22, 23) and is rapidly assuming the role of one of the most important mechanisms of drug resistance in this setting (9). These patients typically are resistant to standard myeloid protocol and thus have a poor prognosis (24). Occasionally, patients are able to achieve remission, but the remission is of short duration (25). Therefore, determining this phenotype is imperative for the institution of appropriate therapy. Review of the patient data in the present study reveals survival differences among the P-glycoprotein strongly positive, weakly positive, and negative patients. Of note, one of the P-glycoprotein weakly positive patients received erythropoietin (Patient 26), as there is a belief that this agent will cause erythroid maturation. A recent report (26), however, has shown that instead of inducing maturation, erythropoietin actually promotes erythroid progenitor survival, thereby promoting growth of the leukemic clone.

Although many of the M6a cases were P-glycoprotein negative (52.3%), the patients with M6b and M6c uniformly demonstrated positivity for P-glycoprotein. The majority of patients (86.7%) with the M6b subtype demonstrated a strongly positive immunohistochemical staining pattern, lack of response to aggressive chemotherapy, and a dismal outcome. Patients with the M6c subtype and a strong reaction to P-glycoprotein fared slightly but not significantly better. The two patients with M6b and a weak P-glycoprotein were the only two patients of this subtype to achieve a brief remission and showed an improved survival (Patients 9 and 15). Similarly, the patients with M6c and a weak immunostain reaction also showed a slight survival advantage.

Most important, as previously demonstrated with other acute myelogenous leukemias (9), P-glycoprotein expression exhibits a statistically significant negative additive effect on response to chemotherapy (remission rate) and survival with unfavorable cytogenetic anomalies. These findings clearly demonstrate the clinical importance of assessment of both the MDR-1 expression and karyotype at the time of diagnosis in this patient population.

In summary, acute erythroleukemia is a stem cell disorder that has demonstrated a high incidence of immunohistochemical positivity for the protein product of MDR-1 at the time of diagnosis. This protein expression demonstrates a negative additive effect with cytogenetic aberrations and most probably contributes to the dismal prognosis in this patient population. Therefore, new chemotherapeutic regimens should be devised on the basis of the specific drug sensitivities of the leukemic cells, combined with p53 gene status and the morphologic subtype of erythroleukemia.