Severe coagulopathy and a high incidence of early fatal hemorrhage characterize acute promyelocytic leukemia [10, 11]. The risk for hemorrhage is exacerbated further by cytotoxic chemotherapy, probably because of rapid cell lysis and release of intracellular procoagulants [12]. Despite the high initial mortality rate, patients with acute promyelocytic leukemia who achieve complete remission appear to have better long-term survival compared with patients who have other types of acute myeloid leukemia [13-19].

Recent studies in China, France, Japan, and the United States showed that all-trans retinoic acid induces complete remission in many patients with acute promyelocytic leukemia [20-24]. Because all-trans retinoic acid initially induces cytodifferentiation rather than immediate lysis of leukemic cells [21, 22], this agent theoretically should decrease the early mortality rate. However, if fewer leukemic cells were eliminated with all-trans retinoic acid than with cytotoxic drugs, such therapy might prove disadvantageous by increasing the proportion of patients who subsequently relapse. To evaluate the role of this drug in an integrated treatment strategy, we conducted a clinical study that used all-trans retinoic acid to induce remission and as maintenance therapy in patients with acute promyelocytic leukemia.

Methods

Top Methods Results Discussion Author & Article Info References

Patients

Our study comprised consecutive patients treated with all-trans retinoic acid during a 2-year period at our institution from June 1990 through June 1992. (A report on the first 11 cases in this series was previously published [22].) We considered patients for this study if they fulfilled morphologic criteria of acute promyelocytic leukemia (M3 or M3-variant) according to the French-American-British classification [25]. Patients with newly diagnosed disease and those who had relapsed from cytotoxic chemotherapy were eligible. We required that women able to bear children have a negative pregnancy test and cautioned them to use effective contraceptive methods. All patients gave signed informed consent, and our center's institutional review board approved the study.

Treatment To Induce Remission

We administered all-trans retinoic acid in two divided oral doses of 45 mg/m² per day (rounded up to the nearest 10 mg) after meals. Exceptions to this regimen occurred on days when patients underwent additional pharmacologic testing, in which case the drug was taken as a single morning dose; also, several patients received the drug in an oil-based suspension through a nasogastric tube because of endotracheal intubation. We withdrew patients who were shown by molecular testing to lack the PML/RAR- rearrangement and treated them with chemotherapy.

Consolidation and Maintenance Treatment Plan

Previous treatment status determined postremission therapy. Patients who were newly diagnosed continued to receive all-trans retinoic acid for approximately 30 days after they achieved complete clinical remission. We then discontinued the drug and administered three courses of consolidation chemotherapy that consisted of idarubicin and cytosine arabinoside. The first chemotherapy cycle was similar in dosage to conventional induction therapy in which idarubicin (12 mg/m² per day given intravenously for 3 days) and cytosine arabinoside (200 mg/m² per day given intravenously for 5 days) were administered concurrently. Subsequent cycles were repeated every 3 to 6 weeks, depending on patient tolerance, using idarubicin (at the same daily dosage for 2 days) and cytosine arabinoside (200 mg/m² per day for 4 days). Five patients received a slightly different consolidation chemotherapy: Three received daunorubicin rather than idarubicin; one refused further treatment after only two of the three planned chemotherapy courses; and one pediatric patient received six cycles of chemotherapy at lower dosages. Two other newly diagnosed patients did not receive chemotherapy (because of advanced age and patient refusal); rather, they received only all-trans retinoic acid. The initial cohort of patients who relapsed from previous chemotherapy-induced remissions were given all-trans retinoic acid therapy unless they were eligible for bone marrow transplant. When we found that remissions maintained by all-trans retinoic acid were brief, a subsequent cohort of previously treated patients received consolidation therapy using a radionuclide-conjugated monoclonal antibody [26].

Management of Leukocytosis, the "Retinoic Acid Syndrome," and Coagulopathy

An early report indicated that patients in whom leukocytosis developed fared poorly [27], and management of this complication evolved during the course of these studies. Although most patients received no specific treatment for leukocytosis, we used leukapheresis for five patients who had rapid increases in their peripheral blood leukocyte counts; intravenous infusions of cytosine arabinoside (100 mg/m² per day for several days) were also administered to five patients. After we recognized a distinctive respiratory distress syndrome as a specific complication of all-trans retinoic acid therapy [28], we gave patients intravenous dexamethasone (10 mg every 12 hours for 3 to 5 days) as prophylactic treatment at the earliest clinical sign of dyspnea. Contrary to a previously routine practice at this center and elsewhere, generally we did not use heparin to treat disseminated intravascular coagulation. Instead, patients with laboratory evidence of coagulopathy received frequent (up to twice daily) transfusions of platelets and fresh frozen plasma to maintain the platelet count at more than 50 x 10⁹/L and the fibrinogen level at more than 1.0 g/L. We reserved heparin for patients with thrombosis or refractory coagulopathy.

Follow-up during the Study

Patients were evaluated each day as inpatients during the initial phase of the study and later as outpatients twice each week until complete remission was achieved. During remission induction, serial assessments were done for blood cell counts, serum biochemical values, and coagulation parameters. In early patients, bone marrow aspiration was done every 5 to 10 days, but this was reduced in later patients. Cytogenetic studies were done using Giemsa or quinacrine banding techniques. The RNA extracted from the mononuclear cell fraction of bone marrow aspirates was analyzed using a reverse transcription polymerase chain reaction assay for the PML/RAR- rearrangement [9]. We observed conventional response criteria [29] and assessed response and toxicity in all patients.

Selection of Historical Controls

We compared newly diagnosed patients who entered this study during a 2-year period with a group of historical control patients. The control group consisted of 80 consecutive patients with newly diagnosed acute promyelocytic leukemia who were treated in protocols at our center during the 15-year period that immediately preceded initiation of this study. Induction chemotherapy in the historical control group consisted of cytosine arabinoside combined with daunorubicin (n = 50), idarubicin (n = 18), or amsacrine (n = 12). The median age of this group was 34 years (range, 15 to 71 years). Other aspects of the control patients were described in two recent reports [13, 30].

Statistical Methods

We estimated the probability of event-free and overall survival using the Kaplan-Meier method.

Results

Top Methods Results Discussion Author & Article Info References

Patient Characteristics

We entered into our study 56 consecutive patients admitted to our center for acute promyelocytic leukemia. Thirty-four patients were newly diagnosed and 22 had relapsed from one or more chemotherapy-induced remissions. The clinical characteristics of these patients are summarized in Table 1. We identified the 15; 17 chromosomal translocation in 51 of 56 patients by cytogenetic study or by molecular analysis of the PML/RAR- rearrangement. Five of the 56 patients had a normal cytogenetic karyotype and no mRNA expression of PML/RAR-; their leukemic cells showed no evidence of maturation with all-trans retinoic acid treatment, and we withdrew them from the study when we obtained the molecular result. Five patients entered the study after treatment for another cancer: three for breast cancer, one for Hodgkin disease (each of whom had previously received cytotoxic chemotherapy), and one for prostate cancer. The median initial leukocyte count in the peripheral blood was 1.9 x 10⁹/L (range, 0.4 to 147.7 x 10⁹/L). Four newly diagnosed patients and three patients who had relapsed had M3-variant morphology. Laboratory findings consistent with disseminated intravascular coagulation were detected in 38 of the 56 patients at the time of entry.

Response

As shown in Table 2, 44 of the 51 patients (86%; 95% CI, 76% to 96%) with cytogenetic or molecular evidence of acute promyelocytic leukemia achieved complete remission with all-trans retinoic acid treatment. In patients with PML/RAR- rearrangements, the response in newly diagnosed patients (26 of 30 cases) was similar to that of previously treated patients (18 of 21 cases). We removed 1 previously treated patient early in the study because of asymptomatic leukocytosis, and she achieved remission with full-dose chemotherapy; however, we classified her as a nonresponder for this study. We interrupted continuous dosing with all-trans retinoic acid during induction in 4 patients who required endotracheal intubation. Although we tried to instill the drug through a nasogastric tube, this was generally unsuccessful and we abandoned this method midway through the study. The shortest duration of all-trans retinoic acid treatment that induced remission was 17 days. The day of therapy on which various response criteria were met is shown in Table 2. For all patients in both the newly diagnosed and previously treated groups, the median time to complete remission by all clinical criteria was 39 days (range, 18 to 78 days). Three of the 5 patients who lacked PML/RAR- rearrangements achieved remission with cytotoxic chemotherapy; all of those responding patients were newly diagnosed.

Duration of Remission and Survival Comparison

In the group of newly diagnosed patients with PML/RAR- rearrangements, 22 of the 26 patients who achieved complete remission received consolidation chemotherapy. Their median relapse-free survival time exceeds 28+ months (range, 1 to 34+ months). The median duration of survival for all newly diagnosed patients exceeds 31+ months (range, 0.4 to 36+ months).

We compared the overall survival of these newly diagnosed patients with that of the historical control patients in several ways. We included all patients who entered this study during the entire 2-year period in the primary analysis, regardless of their final molecular diagnosis and type of induction or consolidation treatment (n = 34; median follow-up, 25+ months), and compared them with the 15-year consecutive series of patients with a morphologic diagnosis of acute promyelocytic leukemia who were treated at this center in protocols that immediately preceded this study (n = 80). (We considered this analysis primary because we presumed that the control group also contained patients who lacked the PML/RAR- rearrangement.) The median duration of first remission and median overall survival time of the control patients were 22 months and 17 months, respectively (follow-up, 95 months). Thus, we found an increase in overall survival Figure 1 for patients who received all-trans retinoic acid for remission induction compared with patients treated only with chemotherapy. The incidence of death during induction in the historical control group was 21% (17 of 80 patients). Although this Figure was somewhat lower than that reported in other series [20, 21], the early mortality rate of newly diagnosed patients treated with all-trans retinoic acid in our study (5 of 30, 17%) was not significantly different from that of the controls.

View larger version (11K): [in this window] [in a new window]   Figure 1. Kaplan-Meier plots of the overall survival of 34 newly diagnosed patients with acute promyelocytic leukemia treated with all-trans retinoic acid (left panel, closed boxes) compared with historical control patients treated with conventional chemotherapy (right panel, closed circles) (n = 80).

A high initial leukocyte count has been the single factor most commonly associated with a poor prognosis in patients with acute promyelocytic leukemia [22, 25], and we found similar results in a multivariate analysis of our historical control patients [30]. To account for possible effects of an imbalance in this variable, we conducted a separate analysis of the survival of patients stratified according to initial leukocyte counts (> or < the median initial count in the control group, 2.4 x 10⁹/L). In this analysis, induction treatment with all-trans retinoic acid still proved superior to conventional chemotherapy (P = 0.014). Less consistently, the presence of the 15;17 chromosomal translocation (identified by conventional cytogenetic analysis) [31, 32] and young age [19-22] have been identified as favorable prognostic features. In the recent analysis of our control group [30], these factors proved nonsignificant for prognosis; however, we compared the survival of the t(15; 17)-positive patients from the control group (n = 24) with that of t(15;17)-positive patients (by cytogenetics) treated with all-trans retinoic acid (n = 23). Overall survival time was still significantly greater for patients who received all-trans retinoic acid (P = 0.03). (A similar result was also noted when the larger group of patients with PML/RAR- molecular rearrangements [n = 30] were analyzed for this comparison.) We did not analyze age comparatively because, unlike the current study, protocols used for the chemotherapy control group specified an upper age limit for entry. The median age of patients in the all-trans retinoic acid group was somewhat greater than the age of control group patients (41 years compared with 34 years), and thus this factor would have been anticipated to bias results in favor of chemotherapy-treated patients.

Duration of Remission in Patients Who Received Only All-trans Retinoic Acid

Thirteen patients received all-trans retinoic acid both to induce remission and for maintenance. Three of these patients were newly diagnosed and did not receive chemotherapy consolidation for reasons of advanced age, refusal, or psychosocial factors, respectively. With one exception (a patient who previously received chemotherapy for breast cancer and who relapsed at 23 months after having received all-trans retinoic acid for 4 months as her only antileukemia treatment), all patients treated and maintained only with all-trans retinoic acid relapsed within 10 months after starting treatment. The median duration of remission for the 13 patients was 3.5 months (range, 1.5 to 23 months). Given this relatively brief remission, consolidation treatment in subsequent patients (n = 8) was changed to a radionuclide-conjugated monoclonal antibody [26], and results from this ongoing study will be reported later.

Response to Re-treatment with All-trans Retinoic Acid

Ten patients (all of whom had previously relapsed from a chemotherapy-induced remission) relapsed again from a subsequent remission while still taking all-trans retinoic acid. These patients were then continued on the drug, usually at an increased dose (90 mg/m² per day) [33]; however, none of them achieved remission. Nine patients who had previously relapsed from a complete remission induced by all-trans retinoic acid were treated again with the drug at some time after they withdrew from retinoid therapy. The median time off all-trans retinoic acid treatment in this group was 6 months (range, 3 to 25 months). Only 3 of these 9 patients (who had been off the drug for 4, 10, and 25 months) achieved a second complete remission with all-trans retinoic acid. Two of the patients died before their response could be fully evaluated (after 9 and 18 days of therapy); four other patients clearly were not responding and we discontinued the drug after 30, 31, 37, and 38 days, respectively.

Leukocytosis, Secondary Leukopenia, and Resolution of Coagulopathy

Most patients began treatment with absolute leukopenia. However, 7 of the 56 patients (13%) had an initial leukocyte count greater than 10.0 x 10 (⁹/L), 6 of whom were in the cohort of previously untreated patients. Unlike the usual experience with cytotoxic drugs, the peripheral blood leukocyte count usually increased with initiation of all-trans retinoic acid treatment. A plot of the median daily leukocyte count for the 51 patients with the PML/RAR- rearrangement is shown in Figure 2. Twenty-seven patients (53%) had total peripheral leukocyte counts greater than 20.0 x 10 (⁹/L) at some time before they achieved complete remission. The median peak leukocyte count was 18.5 x 10⁹/L (range, 4.0 to 176 x 10⁹/L); the median time to the peak leukocyte count was 13 days (range, 2 to 63 days). Examination of the peripheral blood smear frequently showed a population of myeloid cells in intermediate stages of differentiation with nuclear and cytoplasmic vacuolation [34]. After 3 to 5 weeks of treatment, transient leukopenia (<3.0 x 10⁹/L) was commonly observed and occurred in 33 of 46 (72%) evaluable patients after day 14. The median lowest count was 1.8 x 10⁹/L (range, 0.1 to 2.9 x 10⁹/L), which occurred at a median time of 33 days (range, 15 to 44 days). The median peak leukocyte count was somewhat higher in newly diagnosed patients compared with previously treated patients (24.0 compared with 15.0 x 10⁹/L, respectively), but this parameter did not correlate with clinical response.

View larger version (26K): [in this window] [in a new window]   Figure 2. Plot of median total peripheral blood leukocyte count (left axis) of 51 patients with acute promyelocytic leukemia and rearrangements of PML/RAR- treated with all-trans retinoic acid. In addition to the early wave of leukocytosis, a secondary wave of leukopenia is shown between 3 and 5 weeks. (The horizontal dotted line represents the lower limit of the normal range for the peripheral blood leukocyte count. The number of patients who had samples taken on each day is represented in the area curve at the bottom of the Figure forthe right ordinate.).

Several patients were treated specifically for their leukocytosis. Five patients received cytosine arabinoside (100 mg/m² per day for several days), and five had one or more leukaphereses. (One patient was treated with both.) Three of the five patients who received cytosine arabinoside died, all of whom had severe complications of rapid cell lysis (renal failure, exacerbation of coagulopathy, or pulmonary hemorrhage), whereas a fourth had a major, nonfatal intracranial hemorrhage. Two of the five patients who had leukapheresis also died; one of these deaths (from intracranial hemorrhage) was temporally associated with the procedure itself.

Resolution of coagulopathy has been reported as the earliest sign of clinical response to all-trans retinoic acid [35]. Of the 38 patients who had evidence of coagulopathy, 28 had repeated testing sufficient to evaluate the resolution of coagulation abnormalities. Fifteen patients had biochemical evidence that coagulopathy had resolved on or before day 6 of treatment; however, 13 patients took longer than 14 days to resolve, and coagulation abnormalities were still present in 9 patients on or after day 23, although none of these patients required specific therapy for this problem. Seven patients received heparin, 4 for thromboses (three at catheter sites and one in the popliteal vein) and 3 for coagulation abnormalities. As previously noted, 2 patients died of intracranial and pulmonary hemorrhages; 2 others had major nonfatal bleeding from intracranial and gastrointestinal sites.

The "Retinoic Acid Syndrome"

In a previous report [28], we noted that a syndrome characterized by fever, respiratory distress, interstitial pulmonary infiltrates, pleural effusions, and weight gain developed in 9 of the first 35 patients entered into this study. Given the similarity of these symptoms to other clinical events, such as congestive heart failure or pneumonia, we reviewed all cases in this series retrospectively. Of the 56 patients who entered this study, we classified characteristic signs and symptoms of this syndrome as "definitely present" in 13 patients (23%) and "definitely absent" in 33 patients (59%); 10 patients (18%) were thought to have "indeterminate" signs or symptoms. As previously reported [28], this syndrome was also observed in 1 patient who proved to have no PML/RAR- rearrangement. The appearance of this syndrome was not correlated with cell-surface marker expression (CD2, CD10, or CD33), breakpoint location on chromosome 15, level of the initial peripheral blood leukocyte count, or treatment status (for example, newly diagnosed or relapsed from previous chemotherapy) [36] (data not shown). Eight patients required transfer to an intensive care unit for management, and 5 patients died from this syndrome (4 during induction and 1 while in complete remission). As previously noted, the early administration of corticosteroids appeared to abort progression of this syndrome in later patients.

To prevent this syndrome, recent European studies using all-trans retinoic acid have mandated intervention with full doses of cytotoxic chemotherapy based on the rate of increase in the total peripheral blood leukocyte count (for example, 5.0 x 10⁹/L at day 6, 10.0 x 10⁹/L at day 10, and 15.0 x 10⁹/L at day 15) [37, 38]. Therefore, we determined whether these criteria would have predicted the onset of this syndrome in our series. Fully 27 of the 51 patients (53%) with the PML/RAR- rearrangement met these leukocyte criteria, although definite signs of the retinoic acid syndrome were observed in only 23% of patients. Definite evidence of the syndrome developed in only 9 of the 27 patients who met the leukocyte criteria, whereas 6 others had indeterminate symptoms. Conversely, 3 of the 12 definite syndrome cases, along with 4 of the 10 indeterminate cases, never met the aforementioned criteria at any time.

Other Adverse Reactions

The principal adverse effects of all-trans retinoic acid therapy in acute promyelocytic leukemia are listed in Table 3. Headache occurring several hours after drug ingestion was the most common reaction, although most patients rapidly developed a tolerance to this effect. Pseudotumor cerebri, manifested by papilledema, severe headache, and increased cerebrospinal fluid pressure, was documented in three patients who were managed with corticosteroids and lumbar punctures. Dry skin and cheilitis were also common, and two patients had marked scrotal excoriations with ulceration. Musculoskeletal aches and bone pain were observed in 23% of patients. Bone pain was sometimes severe (requiring intravenous narcotics for relief) and tended to be focal rather than diffuse; however, patients also developed tolerance to this symptom and it disappeared despite continued drug treatment. Patients who received corticosteroids for other reasons also reported rapid relief of musculoskeletal pain. Hypertriglyceridemia of various degrees was noted in almost all patients who were serially evaluated (Table 3); however, hypercholesterolemia (more than twice the upper limit of normal) was observed in only one patient. Other than respiratory distress, the only finding that required an interruption in drug therapy was transient hepatic dysfunction, which developed in five patients. The other hepatic or renal biochemical abnormalities noted in Table 3 occurred in an intensive care unit setting with the concomitant administration of multiple drugs.

Discussion

Top Methods Results Discussion Author & Article Info References

Cytotoxic chemotherapy is effective initial treatment for acute promyelocytic leukemia, and most patients (60% to 80%) initially treated with chemotherapy achieve remission [10, 13-17]. However, despite improvements in antibiotics, transfusion therapy, and other aspects of supportive care, the use of cytotoxic drugs to induce remission remains quite toxic, and it fails to cure most patients. Several recent analyses, including that of the control group for this study, have shown that the prognosis of these patients has not improved substantially in the past 15 years after the introduction of anthracyclines and aminoglycoside antibiotics [13, 14, 30].

Agents that induce cytodifferentiation have afforded an opportunity to develop alternative therapeutic strategies. All-trans retinoic acid, which targets the specific molecular lesion in acute promyelocytic leukemia, has great theoretical appeal because it avoids the nonspecific cytotoxicity of conventional drugs along with their exacerbation of coagulopathy in this disease. Initial treatment with all-trans retinoic acid is associated with differentiation of leukemic cells [21, 22]. As patients enter complete remission, morphologically mature cells that retain phenotypic or genotypic evidence of their derivation from the malignant clone can be readily detected [21]. Whereas chemotherapy commonly ablates both leukemic and normal bone marrow cells and causes profound pancytopenia, the usual initial response to all-trans retinoic acid is an increase in the total leukocyte count Figure 2, although the function of these early myeloid cells is not entirely normal [39]. Transient leukopenia is commonly observed after 3 to 5 weeks of treatment, which is probably due to elimination of differentiated leukemic cells and repopulation with cells derived from normal progenitors. Recent experience has not shown a relation between the dose of all-trans retinoic acid and the extent of leukocytosis [40], suggesting that the latter event is inextricably linked to biological response. However, several patients in whom treatment was discontinued early subsequently had complete remission. Similar to observations in certain experimental systems [41], maximum cytodifferentiating effects may be achieved after relatively brief drug exposure, during which sensitive cells are irreversibly programmed for response.

The relatively brief duration of remission in patients treated with all-trans retinoic acid alone suggests that residual leukemic cells escape the drug's effects because of either inherent insensitivity or acquired resistance. No data suggest that ongoing differentiation occurs during prolonged remissions maintained by all-trans retinoic acid. With one exception, all patients who received all-trans retinoic acid as their sole maintenance therapy relapsed in fewer than 10 months, which was similar to findings in other studies [20-24]. As noted recently in a separate analysis of our patients [8], treatment with all-trans retinoic acid could not eradicate molecular evidence of the PML/RAR- rearrangement from patients ostensibly in "complete" clinical remission; however, consolidation chemotherapy routinely converted these assays to negative in most patients, and serially negative tests have been associated with extended remission [8].

In this series, all patients that proved to have the PML/RAR- rearrangement were initially responsive to all-trans retinoic acid. However, the rapid development of resistance limited the use of this agent for extended treatment, and a second remission could not be achieved in any patient who relapsed while taking the drug. Unlike its isomer, 13-cis retinoic acid, plasma concentrations of all-trans retinoic acid decrease markedly with continuous dosing, probably because of increased expression of intracellular retinoic acid-binding proteins and induction of accelerated oxidative catabolism [33, 42-44]. Although pharmacologic resistance to all-trans retinoic acid has proved unexpectedly long-lived, it may disappear after the drug is discontinued; three patients in this series were successfully re-treated after relapses that occurred 4, 10, and 26 months after stopping therapy. Collectively, these data suggest that continuous dosing as maintenance therapy will probably not be effective in eradicating molecular evidence of disease, and relapses that occur during such therapy preclude the opportunity for a second response.

Important limitations apply to the interpretation of studies using historical controls. The control patients were not randomized and they were treated during a 15-year period; thus, various known or unrecognized factors that might have been more evenly distributed by random assignment cannot be discounted. Despite these cautions, several reasons suggest that this comparison is appropriate. First, both the antileukemic therapy (an anthracycline antibiotic plus cytosine arabinoside) and supportive care remained relatively consistent during this period. Analysis of the control group showed no significant differences in either remission incidence or survival when patients were segregated by 5-year treatment periods (for example, 1974 to 1979, 1980 to 1984, and 1985 to 1989) [30]. Moreover, a further analysis comparing the first 30 newly diagnosed patients treated with all-trans retinoic acid in this study compared with the last 30 patients treated with standard chemotherapy at this center showed a similar survival advantage for the all-trans retinoic acid group [45]. The chemotherapy used for all control patients (an anthracycline plus cytosine arabinoside) remains a standard in many studies, including the ongoing multinational intergroup trial. Second, the number of control patients is relatively large for an uncommon disease, and they were consecutively treated at a single center. Although routine anticoagulation was not used in our study, recent analyses have failed to show significant benefits of such treatment [46-48]. Finally, the control patients have fared quite well compared with patients treated in other large series [14-19, 46].

Our results are consistent with preliminary reports from two European studies that compared the use of all-trans retinoic acid with recent or concurrent treatment with chemotherapy to induce remission [37, 38]. The critical difference from our study, however, is that all patients in both European trials received full doses of cytotoxic therapy if the peripheral leukocyte count exceeded specified criteria (see Results). By these criteria, almost three quarters of patients randomized to receive all-trans retinoic acid in the most recent European study actually received additional full-dose chemotherapy to induce remission [38]. In contrast, no patient in our series received full-dose chemotherapy. We found that criteria based on leukocyte counts were not useful for predicting which of our patients would develop the "retinoic acid syndrome," a distinctive complex characterized by dyspnea, fever, and fluid overload [28]. Similar to the incidence in Europe, more than one half of our patients met these leukocyte criteria, yet fewer than one fourth developed symptoms (and these were not always in the same group); further, two of the five patients who died from this complication were already receiving ventilator support by the time they met the leukocyte criteria. The most probable explanation for our observations is that, by definition, the European criteria select patients with an elevated leukocyte count, the single factor that has been correlated most frequently with a poor prognosis in acute promyelocytic leukemia [17, 19, 30]. In this series, most patients with leukocytosis had no symptoms, and our data indicate that, for most patients, specific treatment directed toward reducing the peripheral leukocyte count is not required and may well be hazardous. Early intervention with high doses of corticosteroids has aborted progression of respiratory distress in most recent patients. Therefore, optimal management of leukocytosis must focus solely on prevention of specific complications (primarily hemorrhage) rather than on treatment of the absolute level of the leukocyte count.

Because acute promyelocytic leukemia has been characterized historically by high early mortality rates, perhaps the most surprising result from this study (and contrary to our initial expectations) is the absence of a substantial reduction in the number of early deaths. A similar effect is apparent in both recent European trials [37, 38]. This result appears to be due to the replacement of death caused by intracranial hemorrhage with that resulting from the "retinoic acid syndrome". Recognition of this syndrome and proposals for its management have occurred only recently, and thus subsequent experience may be more positive. Nonetheless, overall survival was still superior to that found in patients treated only with chemotherapy. Together, these data suggest that the initial treatment with all-trans retinoic acid probably yielded a very substantial antileukemic effect, perhaps exceeding that achieved by induction treatment with cytotoxic drugs, which was reflected both early (by a high percent remission) and late (by a reduced risk for relapse). Although additional follow-up is required to confirm these promising intermediate results, this approach to induce remission is clearly associated with significantly less morbidity and lower cost [45]. By comparison, this benefit of reduced morbidity was not observed in the recent European series [38] in which most patients treated with all-trans retinoic acid also received full-dose chemotherapy during induction.

The respiratory distress syndrome has proved an unexpected accompaniment of differentiation therapy in acute promyelocytic leukemia. Assuming management of this problem can be optimized, an important challenge is to determine whether the nonspecific cytotoxicity of conventional anticancer drugs can be avoided altogether, perhaps by using a retinoid suitable for maintenance therapy. Its peculiar pharmacologic properties suggest that all-trans retinoic acid will not be that drug, but analogs that lack these properties may prove more suitable for tests of long-term maintenance or for differentiation therapy in other neoplastic diseases.

Presented in part at the annual meeting of the American Society of Hematology, Anaheim, California, 8 December 1992.