Persistence of Peripheral Blood and Bone Marrow Blasts during Remission Induction in Adult Acute Lymphoblastic Leukemia Confers a Poor Prognosis Depending on Treatment Intensity Free

With modern chemotherapy for the treatment of adults with ALL,² CR rates in excess of 80% have been reported in many series (1, 2, 3, 4). However, a significant proportion of patients eventually relapse, and only 20–45% are eventually cured (5, 6). Therefore, achievement of remission is not by itself sufficient to predict long-term prognosis. Although several pretreatment clinical characteristics, such as older age, high WBC count, and the presence of the Philadelphia chromosome, correlate with a high risk of relapse (1, 7), these are predetermined characteristics that cannot be modified. Identification of events that occur during the course of induction therapy may allow early recognition of patients with a high probability of relapse and could be used to investigate early treatment modifications targeted to improve the CR duration.

It has been hypothesized that even when CR is achieved within 4–6 weeks, a slow rate of leukemia cytoreduction may allow persistence of a large tumor burden, which might allow for the development of drug resistance (8). Previous studies have demonstrated the negative prognostic significance of the persistence of blasts in bone marrow aspirates performed 1–2 weeks after the start of induction chemotherapy in childhood ALL (9, 10, 11). Slow clearance of peripheral blood blasts after induction chemotherapy has also been recognized as a poor prognostic sign in pediatric ALL (12, 13). More recently, molecular evidence of minimal residual disease at the end of the induction chemotherapy was found to have an adverse impact in childhood ALL (14, 15, 16, 17, 18, 19).

Adults with ALL have a better probability of cure when achieving a CR within 4–6 weeks from the start of therapy (1, 4, 7), but the significance of earlier markers of long-term outcome, such as the blast percentage in peripheral blood or bone marrow, and the impact of therapy on these markers, has not been investigated.

In this study, we evaluated the prognostic significance of persistent marrow blasts in the bone marrow 2 weeks after the start of induction therapy (D14BMb) and of peripheral blood blasts 1 week after induction therapy (D7PBb) in adults with newly diagnosed ALL. We also compared the significance of these prognostic factors in two populations treated with chemotherapy regimens of different intensity.

Three hundred sixty five consecutive adults with newly diagnosed ALL treated between 1984 and 1996 were analyzed. Two hundred nineteen were treated with the VAD regimen (1984–1992), and 146 patients were treated with a more dose-intensive regimen of HCVAD (1992–1996). Bone marrow aspirates were obtained on day 14 after the start of chemotherapy in 162 (74%) patients treated with VAD and in 137 patients (94%) treated with HCVAD. Peripheral blood blasts were present at diagnosis in 185 patients (84%) treated with VAD and 119 patients (82%) treated with HCVAD. Persistence of blasts in the peripheral blood 7 days after the start of induction chemotherapy (D7PBb) was evaluated in 143 of 185 (77%) of the VAD-treated patients and in 112 of 119 (94%) of the HCVAD-treated patients.

The VAD regimen has been described previously 1. The HCVAD regimen is summarized in Table 1.

Early death and failure to achieve remission were considered from the time of start of treatment to date of the event. Events for EFS included failure to achieve CR, relapse, or death from any cause. Survival was determined from the date of start of induction chemotherapy until date of death. Remission duration was calculated from the time of documented CR until evidence of relapse. Categorical variables were compared among patients with and without persistent D7PBb and D14BMb using the χ² or Fisher’s exact test and continuous variables by the Mann-Whitney test (20). Curves were plotted by the method of Kaplan and Meier (21) and compared by the log-rank test (22). EFS, according to the different prognostic factors, was estimated with the Cox proportional hazard model (20). Relevant prognostic factors were first investigated by univariate analysis. A stepwise multivariate regression then identified the significant independent prognostic variables.

The clinical characteristics of the study population are shown in Table 2. There were no significant differences between patients treated with the VAD regimen and those treated with the HCVAD regimen. Two hundred and ninety three of the 365 patients analyzed (80%) achieved a CR, including 160 of 219 (73%) treated with VAD and 133 of 146 (91%) treated with HCVAD (P < 0.0001). Median CR duration was 98 weeks, with 41% with CCR at 3 years. There was a trend in favor of patients treated with HCVAD (median, 123 weeks; 48% at 3 years) compared with those treated with VAD (median, 85 weeks; 41% at 3 years; P = 0.09). The median EFS was 58 weeks, with 29% alive and free of disease at 3 years. EFS was significantly better for patients treated with the HCVAD regimen (median, 81 weeks; 40% at 3 years) compared with those treated with VAD (median, 38 weeks; 23% at 3 years; P = 0.00003). The median overall survival was 84 weeks, with 33% alive at 3 years, with significantly better results for the HCVAD group (median, 117 weeks; 46% at 3 years) compared with the VAD group (median, 72 weeks. 27% at 3 years; P = 0.0004)

After 7 days from the start of induction chemotherapy, 117 of 255 (46%) patients with peripheral blood blasts at diagnosis had residual blasts by day seven. These included 79 of 143 (55%) treated with VAD and 38 of 112 (34%) treated with HCVAD (P = 0.001). For patients treated with VAD, an initial WBC count ≥30 × 10⁹/liter was associated with persistent D7PBb (P = 0.005). There were no significant differences in the pretreatment characteristics of patients with or without D7PBb among those treated with HCVAD. Patients with no D7PBb had a higher CR rate (126 of 138; 91%) than those with D7PBb (88 of 117; 75%; P = 0.0005). This was mostly due to a significant difference among patients treated with VAD, where 57 of 64 (89%) patients who had no D7PBb achieved a CR, compared with 55 of 79 (70%) with D7PBb (P = 0.005). There was no difference in CR rate among patients with or without D7PBb after treatment with HCVAD (87% versus 93%; P = 0.26). However, persistence of D7PBb was associated with significantly worse EFS at 3 years in patients treated with HCVAD (25% versus 55%; P = 0.003; Fig. 1,B) but only a trend among those treated with VAD (18% versus 28%; P = 0.12; Fig. 1,A). A similar association was observed for CR duration, with inferior CCR at 3 years among HCVAD patients (36% versus 55%; P = 0.02) and a nonstatistical difference among VAD patients (28% versus 42%; P = 0.28). Survival was also significantly worse for HCVAD patients with persistent D7PBb compared with those without D7PBb (27% versus 60%; P = 0.001; Fig. 2,B), but there was no significant difference among VAD-treated patients (23% versus 34%; P = 0.4; Fig. 2 A).

Persistent D14BMb were found in 131 of 299 patients (44%) in whom a bone marrow aspiration was obtained on day 14, including 80 of 162 patients (45%) in the VAD group and 51 of 137 (37%) in the HCVAD group (P = 0.046). Patients with persistence of D14BMb were more frequently Philadelphia chromosome-positive than those without D14BMb, both in the VAD group (23% versus 9%, respectively; P = 0.02) and in the HCVAD group (20% versus 8%; P = 0.06). Also, an initial WBC >30 × 10⁹/liter was more common among patients with D14BMb than among those without D14BMb, both in the HCVAD group (35% versus 23%, P = 0.016) and in the VAD group (34% versus 23%, P = 0.18). No other pretreatment clinical characteristics were associated with D14BMb. There was a strong association between D14BMb and D7PBb, both among patients treated with HCVAD (P = 0.0005) and those treated with VAD (P = 0.0001). Persistent D14BMb was associated with a significantly lower CR rate in the VAD group (60% versus 90%; P < 0.0001) but not among those treated with HCVAD (92% versus 92%; P = 0.95). Persistence of D14BMb was associated with a significantly worse 3-year EFS in the VAD group (37% versus 18%; P = 0.00006; Fig. 3,A) but only a trend among the HCVAD-treated patients (43% versus 31%; P = 0.19; Fig. 3,B). CCR at 3 years was also negatively influenced by D14BMb among VAD patients (44% versus 27%; P = 0.02) with a trend among HCVAD patients (53% versus 39%; P = 0.18). Survival was also worse for VAD-treated patients with persistent D14BMb (18% versus 39%; P = 0.01; Fig. 4,A) but not for HCVAD-treated patients (43% versus 49%; P = 0.61; Fig. 4 B).

Other characteristics with recognized prognostic significance were associated with outcome in a univariate analysis. CR rate correlated with age, FAB group (worse for L3), cytogenetics, serum β₂-microglobulin, albumin and bilirubin levels, platelet count, D7PBb, D14BMb, and treatment (VAD versus HCVAD). EFS was associated with cytogenetics, FAB subgroup, age, performance status, serum albumin levels, hemoglobin, platelet count, D7PBb, D14BMb, and treatment. Survival was associated with FAB subgroup, cytogenetics, age, performance status, albumin levels, platelet count, D14BMb, and treatment.

We performed a multivariate analysis to investigate whether the prognostic significance of D7PBb and D14BMb was independent of other prognostic factors (Table 3). The characteristics associated with CR rate in a multivariate analysis included treatment (VAD versus HCVAD), FAB subgroup (L3 versus others), presence of D14BMb, age, and total bilirubin. However, the significance of D14BMb varied with treatment; it had an independent prognostic relevance for patients treated with VAD (P = 0.0002) but not for those treated with HCVAD (P = 0.82).

Characteristics associated with EFS in a multivariate analysis included FAB, cytogenetics (Philadelphia-positive versus others), D7PBb, D14BMb, age, performance status, and platelet count. The significance of D7PBb and D14BMb also was dependent on therapy. For patients treated with VAD, D14BMb (P = 0.001) but not D7PBb (P = 0.32) was an independent prognostic factor. The opposite was true for patients treated with HCVAD (D7PBb, P = 0.03; D14BMb, P = 0.26).

A similar association was observed for survival where D7PBb (P = 0.02) was an independent prognostic factor in the total group, together with treatment, cytogenetics, age, performance status, and platelet count; D14BMb was not associated with survival (P = 0.25). However, for patients treated with VAD, D14BMb was associated with survival (P = 0.06), with no association noted for D7PBb (P = 0.40). D7PBb was an independent prognostic factor for patients treated with HCVAD (P = 0.005), but D14BMb had no association with survival (P = 0.96).

Patients with persistent D7PBb can still be separated in two prognostic groups by D14BMb. Among the 104 patients who had D7PBb, the 3-year EFS was 39% for those without D14BMb compared with 12% for patients with D14BMb (P = 0.0001). This difference was mostly in 68 such patients treated with the VAD regimen, where the 3-year EFS was 10 and 42% (P = 0.0001) for those with and without D14BMb, respectively. Only a trend was present among patients treated with HCVAD (12% versus 21%; P = 0.58).

Several characteristics with prognostic significance in adult ALL have been identified by multivariate analysis in previous studies (1, 7, 23). These factors usually represent characteristics of the host (e.g., age, performance status, and others) or the disease (e.g., cytogenetics, FAB subgroup, WBC count, and others) present at the time of diagnosis. Early response to therapy (i.e., achieving a CR within 4–6 weeks from the start of therapy or after one course of induction chemotherapy) has been reported as a prognostic factor in some series (1, 4, 7).

In this report, we identify early disappearance of blasts from the peripheral blood or bone marrow as prognostic factors for survival and EFS in adult ALL. Also, this is the first report identifying differential prognostic implications of D7PBb and D14BMb according to therapy, where D14BMb was associated with prognosis in patients treated with the VAD regimen and D7PBb was most significant for patients treated with the HCVAD regimen.

Early disappearance of blasts from the bone marrow has been identified as a favorable prognostic factor in childhood ALL. Jacquillat et al. (24) reported a significantly shorter remission duration for patients requiring more than two weekly doses of vincristine and prednisone to clear bone marrow blasts. Frei and Sallan 8 reported a higher relapse rate (62%) among patients with <50% cytoreduction by day 5, compared with those with more complete cytoreduction (relapse rate, 10%). Large cooperative pediatric trials have further expanded these observations. Miller et al. (25) identified persistence of blasts on day 14 bone marrow as an adverse prognostic factor for CR and disease-free survival by multivariate analysis. These results were later confirmed in a report of the CCG-160 protocol series (26), including 2987 children, with 6-year event-free survival rates of 63% for patients with <5% bone marrow blasts by day 14, compared with 44 and 25% for those with 5–25% and >25%, respectively (P = 0.0001). By multivariate analysis, persistence of blasts in day 14 bone marrow ranked fourth after age, WBC count, and gender (26). Most interestingly, persistence of blasts in day 14 bone marrow was the single most important prognostic factor for patients alive and disease-free 2 years after achieving remission, suggesting very late effects of early response to chemotherapy (26). Other studies have found the association of persistent bone marrow blasts with adverse outcome by univariate but not multivariate analysis (11). Persistence of bone marrow blasts on day 7 was also significant, where patients with ≥25% blasts had a decreased probability of 3-year EFS (47%), although there was no difference between patients with <5% (77% 3-year EFS) and those with 5–24% blasts (80% 3-year EFS). In fact, when bone marrow is evaluated at both days 7 and 14, a complementary prognostic evaluation can be obtained; patients with <5% blasts by day 7 have the best 5-year EFS rate (80%), followed by those who take until day 14 to clear blasts from the bone marrow (60%), and only 24% for those with persistent blasts by day 14 (9). Clearance of peripheral blood blasts seems to be important also. Rautonen et al. (13) reported a 5-year EFS of 60% for patients with peripheral blood blast clearance by day 10, compared with 36% when blasts cleared between days 11 and 13, and 17% if they cleared only after 13 days (P = 0.003). Similar results were reported by Gajjar et al. 12, and both studies showed peripheral blood clearance as an independent prognostic factor by multivariate analysis. No correlative studies have been reported to assess the correlation between bone marrow and peripheral blood blasts.

The significance of early response to therapy and prognosis has not been fully investigated in adult ALL. Patients who achieve a CR within 4 to 6 weeks from start of induction chemotherapy have a better outcome (1, 4, 7). Keating et al. (27) reported a significant reduction in expected EFS for acute leukemia patients in whom the number of leukemic cells had not been reduced to half by day 5 of treatment in adults with various acute leukemias. Sebban et al. (28) reported a negative impact on survival and disease-free survival of persistent blasts in bone marrow by day 15. These authors suggested that these data could be used on treatment planning. Our results represent the first study to demonstrate the prognostic implications of persistent peripheral blood and add to the observation of Sebban et al. (28) on persistent bone marrow blasts in adult ALL. Although D7PBb and D14BMb were associated with other clinical characteristics with prognostic significance in ALL, multivariate analysis demonstrated an independent association of D7PBb (P = 0.02) and D14BMb (P = 0.007) with EFS. Although some of the effect on EFS was due to early events (i.e., CR, early relapse), as reported by Miller et al. (26), a significant portion of the effect of early peripheral blood or bone marrow clearance was due to an impact on later events (Figs. 1,2,3,4). This is shown by the fact that among patients alive and in CR for ≥1 year, the EFS was 66% for patients without D14BMb, compared with 44% for those with D14BMb (P = 0.01). Similar difference in EFS was observed among this group for patients comparing those with and without D7PBb (68 and 44%, respectively; P = 0.01).

Certain therapeutic interventions can change the significance of conventionally accepted prognostic factors. This has been the case for example with immunophenotype, where patients with a T-cell disease were historically considered to have an adverse prognosis (29). With treatment protocols including early administration of cyclophosphamide and cytarabine during induction, patients with T-cell immunophenotype are now considered to have a favorable prognosis in many series (7, 23). The prognoses of adults with mature B-cell ALL have also improved significantly with short-term intensive therapy (30, 31, 32).

In this study, the prognostic implications of D7PBb and D14BMb varied with protocols of different intensity. With the less effective VAD regimen, approximately half of the patients had persistent D7PBb (n = 82; 57%) or D14BMb (n = 80; 49%). D14BMb was a significant and independent prognostic factor for CR (P = 0.0002) and EFS (P = 0.001). For these, patients however, the significance of D7PBb was lost in a multivariate analysis. These results are in fact almost identical to those reported by Sebban et al. (28). With the more effective HCVAD regimen, only 38 patients (34%) had D7PBb and 51 (37%) had D14BMb. For these patients, D7PBb was an independent prognostic factor for EFS (P = 0.03) and overall survival (P = 0.005) but not for CR, most likely due to the high CR rate among these patients (91%). In contrast, D14BMb had no independent prognostic implications among these patients, suggesting that with a more intensive treatment program, where the first course of induction chemotherapy has been completed by day 14 (HCVAD), an earlier clearing of leukemia (i.e., by day 7) occurs more frequently, but those patients who have not cleared the leukemic cells from peripheral blood represent a group with very poor prognosis. Steinherz et al. (9) also found an earlier clearance when a more intensive five-drug regimen was used, increasing the proportion of patients who achieved a M1 bone marrow by day 7 from 56 to 68%. However, in contrast to our results, this did not translate into improved EFS.

Even among patients with persistent D7PBb, clearance of D14BMb gave a survival advantage, mostly for patients treated with a less intensive program. Steinherz et al. (9) identified a group of children with persistent bone marrow blasts on day 7 who cleared them by day 14 (i.e., intermediate response group). These patients had a 5-year EFS of 60%, compared with 80% for those with an early response (i.e., M1 bone marrow by day 7) and 24% for those with persistent blasts on day 14. These observations open the possibility of therapeutic interventions when a slow responder is identified by day 7, where the day 14 response can still make an impact on long-term outcome.

We conclude that early clearance from peripheral blood and/or bone marrow has prognostic implications among adults with ALL. The significance of these prognostic indicators, however, varies according to the treatment used. Because these early indicators have an impact on late events, early therapeutic interventions in response to these observations might improve the long-term prognosis of adult ALL. This hypothesis needs to be tested in prospective trials.