Progressive hyperleukocytosis is a relevant predictive marker for differentiation syndrome, early death, and subsequent relapse in acute promyelocytic leukemia

Acute promyelocytic leukemia (APL) is generally held to have favorable risk, but we have observed a high incidence of early deaths caused by fatal bleeding and differentiation syndrome (DS). We retrospectively analyzed 259 APL patients from 2002 to 2014 who all received all-trans retinoic acid (ATRA) with the support of sufficient transfusions, followed by 4 days of idarubicin. High-risk status was determined as a diagnostic leukocyte count (WBCdx) >10 × 109/L (Sanz criteria). For patients with hyperleukocytosis, we sometimes conducted leukapheresis and also used hydroxyurea and prophylactic dexamethasone. Because we frequently observed patient fatalities from progressive hyperleukocytosis, we also checked the maximum leukocyte count (WBCmax) and stratified patients by their incremental ratios. The 8-week cumulative incidence of early death and DS was 13.5% and 17.8%, respectively. We found that WBCmax correlated better with early death and DS, even in the low-risk group, than WBCdx. Among the patients with WBCdx <10 × 109/L, a WBCmax >43 × 109/L correlated with early death (26.7%) and DS (40.0%). Also, having a WBCdx of 10 to 43 × 109/La that increased to a WBCmax >43 × 109/L correlated with increased early death (33.3%). The multivariate analysis revealed that a WBCmax >43 × 109/L correlated significantly with both early death and DS.


Discussion
In this study, we have found good survival outcomes from our analysis of 259 APL patients treated with ATRA and anthracycline-based chemotherapy based on the AIDA protocol 1,18,19 . As we experienced, except for the 35 (13.5%) patients who died within 8 weeks, NRM was low, and most long-term failures were caused by APL relapses: 5-year NRM was 1.2% and 5-year CIR was 17.1%. Most early deaths were caused by significant bleeding complications, including intracranial hemorrhage and alveolar hemorrhage, followed by multiorgan failure.
We also identified the cumulative incidence of DS, which occurred in 46 patients (17.8%). Of those, severe DS was observed in 9 (19.5%), and 4 of them died of progressive organ failure caused by the severe DS. This value www.nature.com/scientificreports www.nature.com/scientificreports/ is lower than in a previous large cohort study 13 ; most of our DS cases were safely manageable with intravenous dexamethasone and immediate cessation of ATRA. However, we might have missed some patients with severe DS that was misdiagnosed as a bleeding complication, such as alveolar hemorrhage. Of the 35 early deaths in our cohort, 16 patients suffered from alveolar hemorrhage.
Sanz-risk scoring suggests that the leukocyte count prior to therapy is the prognostic factor most relevant to relapse and survival outcomes 17,20 . Although little was previously reported about factors predictive of early death and DS, a study by Vahdat et al. showed that peak leukocyte count correlated with DS 16 , and some studies have suggested that an abnormal creatinine level, male sex, old age, and hyperleukocytosis at diagnosis are related to early death 14,17 . However, no factors have been consistently shown to affect early complications.
We stratified patients into groups by their diagnostic leukocyte count (WBCdx), leukocyte increment ratio, and maximum leukocyte count (WBCmax). Classic Sanz-risk scoring suggests that a WBCdx higher than 10 × 10 9 /L is a significant cut-off for long-term outcomes. Our ROC analysis in this study found that both a WBCdx and WBCmax higher than 43 × 10 9 /L correlated significantly with early death and DS. Overall, we first identified changing outcomes according to 3 levels of leukocyte increment beginning with a WBCdx lower than 10 × 10 9 /L. Progression to both WBCmax 10 to 43 × 10 9 /L and WBCmax >43 × 10 9 /L correlated with a significantly higher incidence of DS and early death compared with patients with a WBCmax <10 × 10 9 /L. In addition, the 2 subgroups with progressive hyperleukocytosis showed significantly higher CIR rates and inferior OS even though they had low leukocyte counts at diagnosis. This might be the first evidence that progressive leukocytosis (as reflected by WBCmax) is a more relevant factor in predicting early and long-term survival outcomes than the leukocyte count prior to therapy. At this point, we question whether ATRA or ATO should be started in patients with hyperleukocytosis and wonder whether cytoreduction or leukapheresis is more important for patients with hyperleukocytosis to reduce the rates of DS and early death. However, no definite guidelines discuss those issues because high-risk APL patients are rare, and individual differences make it difficult to draw definite conclusions for standard management strategies. Nonetheless, the role of prophylactic dexamethasone should be studied in the near future.
The multivariate analysis also revealed that old age correlated with early death, and a low ATIII level correlated with DS. For ATIII, our further analysis revealed that ATIII >100% correlated with significantly lower incidence of DS, even in patients with WBCmax >43 × 10 9 /L. Because we can support ATIII in patients using disseminated intravascular coagulation and a supportive management strategy, ATIII and transfusions could be adjusted in future prospective trials. In old age, especially for people with comorbidity, intensive chemotherapy might be toxic, causing bleeding complications and sepsis. Therefore, using ATO plus ATRA could be an important alternative for elderly APL patients 8,10,11,15 . A randomized phase III trial demonstrated that ATRA plus ATO is at least non-inferior and could be superior to ATRA plus chemotherapy in low-to intermediate-risk APL 8 . In high-risk APL, a more tailored therapy using ATO and attenuating conventional chemotherapy is needed to reduce the early death rate in elderly patients 21 .
Although our current results are from a retrospective study, all enrolled patients were treated with the same management strategy for a long period, which could minimize possible biases. Unfortunately, because the standard treatment for APL has changed to a combination therapy using ATO, these results might no longer be applicable, so our results need further validation. In conclusion, our data show that a WBCmax indicating progressive leukocytosis correlated better with early death and DS than the Sanz-risk criteria. The role of dexamethasone prophylaxis, antithrombin III, and a cytoreduction strategy should be evaluated in specific patient subsets to reduce early events in APL. In addition, being in the high-increment WBCmax group also correlated with a high risk of relapse, so early intervention with close monitoring and preventive management might improve survival outcomes 23 .

Methods
Study population. We initially identified 259 adult APL patients with a median age of 42 years (range; 15-72 years) from 2002 to 2014. We confirmed APL with a chromosomal analysis and additional PML-RARA reverse transcriptase polymerase chain reaction (RT-PCR), followed by reverse transcription quantitative PCR (RT-qPCR). To detect the karyotypes, at least 20 metaphases from BM cells were analyzed by the GTG banding method after 24 or 48 h of unsynchronized culture, and the International System for Cytogenetic Nomenclature was used as a guideline for classification 24 . We also identified the presence of additional chromosomal abnormalities. The purpose and experimental protocols of this research were approved by the Institutional Review Treatment strategy and supportive care. Except for 1 woman, all patients received ATRA (45 mg/ m 2 /day) divided into 2 daily doses immediately upon suspicion of APL or with morphological evidence from a BM aspiration study. We performed leukapheresis whenever leukocyte counts exceeded 50 × 10 9 /L, and some patients with hyperleukocytosis were treated with hydroxyurea, cytarabine, and prophylactic dexamethasone. Available blood products were vigorously administered with frequent monitoring. The target level for transfusion was higher than 50 × 10 9 /L for platelets, higher than 120 mg/dL for fibrinogen, higher than 70% for prothrombin time, and higher than 70% for ATIII. Most patients were treated with induction chemotherapy using idarubicin (12 mg/m 2 , days 1, 3, 5, 7) based on the AIDA protocol, as previously reported 1,18,19 , but some patients (n = 13) with serious infectious complications or comorbidity were treated with a compassionate program of ATO, and some (n = 26) were treated with ATRA alone. After achievement of hematological CR, all patients received the same consolidation chemotherapy consisting of three courses in combination with 15 days of ATRA as follows -course 1, idarubicin (7 mg/m 2 , days 1-4); course 2, mitoxantrone (10 mg/m 2 , days 1-4); and course 3, idarubicin (12 mg/m 2 , day 1-2) -which was based on the LPA99 protocol from PETHEMA 6,25 . After completion of consolidation, patients received the following maintenance therapy for 2 years: 6-mercaptopurine (50 mg/m 2 /day) and ATRA for 15 days every 3 months. Our maintenance was a modified regimen (by exclusion of methotrexate) based on a protocol previously reported 6,19,25 . When DS occurred, we promptly discontinued ATRA and started dexamethasone (10 mg twice daily), with strict control of volume overloading using diuretics or renal replacement therapy. Molecular studies. The molecular studies were performed from the BM samples collected at diagnosis, 1 month after induction and consolidation chemotherapies, and every 3 months after maintenance treatment. PML-RARA, the representative marker for measurable residual disease, was detected by a multiplex RT-PCR screening assay using a HemaVision kit (DNA Technology, Aarthus, Denmark), and quantification was performed using the RT-qPCR method (Real-Q PML-RARA quantification kit, Biosewoom, Korea) with a www.nature.com/scientificreports www.nature.com/scientificreports/ sensitivity of 5.0 log (10 −5 ). The RQ-PCR level represented the ratios of PML-RARA normalized to the expression of the reference gene, ABL1 (1.0 × 10 4 ). The FLT3 mutation was detected using multiplex allele-specific RT-PCR (ABSOULTE FLT3 TKD/ITD RT-PCR, Biosewoom, Korea). Because those molecular studies became available in 2006, we could not obtain molecular data for patients treated before that.
Primary endpoint and definitions. The primary endpoint of the current study was to determine predictive factors that affect early complications, early death, and DS, in APL patients. Therefore, we tried to identify the prognostic value of progressive hyperleukocytosis after ATRA administration during initial treatment. We checked the diagnostic level (WBCdx) and maximum level (WBCmax) of the leukocyte counts during the initial treatment period and identified subgroups with progressive hyperleukocytosis, i.e., those with a low WBCdx <10 × 10 9 /L whose WBCmax was significantly higher than that. The definition of early death was all patients who died within 8 weeks of treatment. DS was diagnosed as the presence of at least 2 of the following -dyspnea, unexplained fever, sudden weight gain greater than 5 kg, unexplained hypotension, acute renal failure, or a chest radiograph demonstrating pulmonary infiltration or pleuropericardial effusion 4,13 . Any patient showing 4 or more of those features was considered to have severe DS, and those with fewer than 4 were classified as having moderate DS.  Table 2. Multivariate analysis of affecting factors for early death and differentiation syndrome. Abbreviation: CIR, cumulative incidence of relapse; HR, hazard ratio; FLT3, Fms-like tyrosine kinase 3; ITD, internal tandem duplication; TKD, tyrosine kinase domain; BCR, breakpoint cluster region; ATRA, all-trans retinoic acid; ATIII, Antithrombin III.