Drug-Induced Liver Injury during Consolidation Therapy in Childhood Acute Lymphoblastic Leukemia as Assessed for Causality Using the Updated RUCAM

Purpose The presence of serious toxicities is a major problem in the treatment of childhood acute lymphoblastic leukemia (ALL). The objective of this research is to evaluate drug-induced liver injury (DILI) during consolidation therapy in childhood ALL. Methods Clinical data of pediatric patients who received consolidation therapy between August 2012 and July 2018 were collected. Characteristics (incidences and patterns) of DILI at different stratifications were determined. Risks of DILI were evaluated using binary logistic regression analysis. Drug causality assessment was carried out by the updated Roussel Uclaf Causality Assessment Method (RUCAM). Results Patients with high risk (HR) and standard risk (SR)/intermediate risk (IR) received 270 and 1539 courses of consolidation therapy, respectively; among these courses, 15 (5.6%) and 38 (2.5%) developed DILI. The occurrences of DILI in SR/IR patients were primarily associated with age (≤5.2 years), treatment course (≥5), and baseline serum parameters before treatment (cystatin C > 0.79 mg/L, albumin ≤45 g/L, and gamma-glutamyl transpeptidase (GGT) > 17 U/L). The ROC curve generated using the parameters assigned to specific values achieved an area under the curve (AUC) of 0.846 (95% CI 0.827–0.863) with a cutoff value of 3, and the sensitivity and specificity were 94.7% and 62.3%, respectively. For HR patients, a decrease in baseline albumin and elevation of baseline liver enzymes (GGT and aspartate aminotransferase) were observed in DILI cases compared with the non-DILI subjects. In the SR/IR group with DILI, the causality gradings for high-dose methotrexate (HD-MTX) were highly probable in 5 (13.2%) cases, probable in 31 (81.6%) cases, and possible in 2 (5.3%) cases. Among the DILI cases in HR-1, HR-2, and HR-3 groups, high causality gradings (probable + highly probable) were detected in “100% of HD-MTX + 57% of high-dose cytarabine (HD-Ara-C),” “100% of HD-MTX + 20% of pegylated asparaginase (PEG-ASP),” and “100% of HD-Ara-C + 33.3% of PEG-ASP,” respectively. Conclusion Incidence of DILI in HR patients was significantly higher than that in SR/IR patients. A number of potential risk factors were identified, among which the preexisting liver conditions were suggested as shared risk factors in all stratification groups. HD-MTX, HD-Ara-C, and PEG-ASP were the main causative agents of DILI. The knowledge generated from this study will be helpful for understanding characteristics of DILI during consolidation treatment in childhood ALL.


Introduction
Acute lymphoblastic leukemia (ALL) is the most common malignancy in children [1]. e treatment of childhood ALL usually consists of following phases: induction, early intensification, consolidation, delayed intensification, and maintenance [2]. During the consolidation phase, patients often receive multiple chemotherapeutic agents, such as 6-mercaptopurine (6-MP), methotrexate (MTX), asparaginase (ASP), cyclophosphamide (CTX), etoposide (VP- 16), and cytarabine (Ara-C). ese chemotherapeutic drugs have a variety of adverse effects, especially when used at high doses or administered in combination. As a result, toxicities associated with treatment are the main obstacles in consolidation therapy of childhood ALL [2,3]. e development of severe toxicities secondary to chemotherapy not only increases the cost of treatment but also delays scheduled therapy and jeopardizes the cure rate for these patients [4]. Among those toxicities, hepatotoxicity is one of the most important barriers during ALL chemotherapy. However, to our knowledge, overall hepatotoxicity during consolidation therapy, especially high-dose MTX (HD-MTX)-based polychemotherapy, has not been fully elucidated in Chinese pediatric patients. In the present study, we characterized the incidences, types, risk factors, and causality assessment of drug-induced liver injury (DILI) during consolidation therapy in childhood ALL.

Study Design.
We retrospectively collected data of pediatric ALL patients who received consolidation therapy at the First Affiliated Hospital of Guangxi Medical University from August 2012 to July 2018. In total, there were 1921 consecutive courses of consolidation therapy during the study period. After excluding those with incomplete data, 1809 courses were included in the analysis ( Figure 1). Incidences and patterns of DILI were determined. Risks at various stratifications were evaluated, and drug causality assessment was performed. Ethical approval was obtained from the Ethical Review Committee of the First Affiliated Hospital of Guangxi Medical University (2021(KY-E-111)).

Patients.
Risk stratification (standard risk, SR; intermediate risk, IR; and high risk, HR) of patients were defined based on their clinical features (e.g., age, initial leukocytosis, immunophenotype, and cytogenetics) and early prednisone treatment response [5]. e treatment of consolidation was conducted at about 2 weeks after the end of early intensification chemotherapy according to the Chinese Children Leukemia Group (CCLG) 2008 protocol [5].

Treatments.
For SR and IR patients, they received 2 000 and 5 000 mg per meter squared (mg/m 2 ) body surface area (BSA) of HD-MTX as a 24-hour infusion, respectively. is process was repeated 4∼6 times with an interval of at least 14 days. 25 mg/m 2 /d of 6-MP was taken orally on empty stomach before bedtime. A single intrathecal injection of MTX for SR patients and triple intrathecal therapy (TIT) composed of MTX, Ara-C, and dexamethasone (DXM) for IR patients were administered at 2 hours after HD-MTX infusion. 15 mg/m 2 of leucovorin (LV) was given at 42 h after the start of MTX infusion, and the dosage of LV at 48 h and later (every 6 h) were adjusted based on MTX plasma concentration.

Data Collection.
We retrospectively collected information from the hospital information system (HIS). Parameter selection criteria for developing models were based on the resources available from the HIS and existing literature. ese parameters include patients' demographic features (e.g., age, sex, ethnicity, body mass index (BMI), BSA, and courses), biochemical parameters (e.g., renal function, liver function, and blood routine), and medication information (e.g., drug doses, administration of potential hepatotoxic agents, and prophylactic use of hepatoprotective drugs). e laboratory tests for liver function were performed as baseline measurements before each course of chemotherapy and were conducted at least once during treatment based on the patients' status. e median detection frequency of liver function after initiating chemotherapy was 1 (range, [1][2][3][4][5][6][7][8][9][10][11][12][13]. e median detection time of liver function was 3 days (range, 1-25 days) after the start of chemotherapy.

Definition and Pattern of DILI.
DILI was detected by serum biochemical criteria, which include alanine aminotransferase (ALT), alkaline phosphatase (ALP), and bilirubin Consecutive courses of consolidation therapy for childhood ALL N=1921 Courses enrolled for data analysis of DILI N=1809 Courses excluded N=112 -no liver biochemical tests available before and/or during treatment Figure 1: Flow chart of courses inclusion. ALL, acute lymphoblastic leukemia; DILI, drug-induced liver injury. [6]. Here, subjects were defined as having DILI if they met one of the following criteria: (1) ALT ≥5 × upper limit of normal (ULN), (2) ALP ≥2 × ULN, or (3) ALT ≥3 × ULN and simultaneous bilirubin >2 × ULN. If the liver biochemistry was abnormally elevated before consolidation therapy, ULN was replaced by the baseline value. e pattern of DILI was identified using the R value, where R � (ALT/ ULN)/(ALP/ULN). DILI was considered "hepatocellular" when R ≥ 5, "cholestatic" when R ≤ 2, and "mixed" when 2 < R < 5. e ULN of ALT was 60 U/L for boys and 45 U/L for girls. e ULN of total bilirubin was 20.5 μmol/L. e ULN of ALP was 500 U/L (age <2 years) and 750 U/L (2-14 years). For children aged >14 years, the ULN of ALP was 125 U/L for boys and 100 U/L for girls.

Statistical Analysis.
As the chemotherapy regimens in SR and IR patients were roughly similar, data of SR and IR groups were merged for statistical analysis. Continuous variables were presented as mean ± standard deviations (SD), and categorical data were expressed as numbers (percentages). e SPSS 20.0 software (SPSS Inc., Chicago, IL, USA) was used for the statistical analysis of the data. For the SR/IR patients, univariate analysis was conducted first and variables with a p value of less than 0.20 were included in binary logistic regression analysis using a Forward LR method. Cutoff values of the identified independent risk factors were determined by the receiver operating characteristic (ROC) curve. e independent factors were assigned values of 0, 1, and 2 based on the rounded regression coefficient (B) and their role in the development of DILI. e area under the curve (AUC), sensitivity, and specificity of the final model were assessed using the ROC curve. For the HR patients, only univariate analysis was performed due to the limited number of DILI cases. e updated Roussel Uclaf Causality Assessment Method (RUCAM) was used to determine the causal relationship between liver injury and suspected drugs [6].

Characteristics of DILI.
A total of 1809 eligible courses involving 463 patients, 293 boys and 170 girls, were enrolled for analysis in the present study. ere were 49 patients in the HR group and 414 patients in the SR/IR groups who underwent 270 and 1539 courses of consolidation therapy, respectively. e median age at consolidation therapy was 6.4 years (range, 1.1-15.7 years). e patients were treated with a median number of 3 courses (range, 1-9 courses). e median time to first detection of DILI after the start of chemotherapy was 3 days (range, 3-8 days). e incidences of DILI in HR-1 and total HR (HR-1 + HR-2 + HR-3)courses were 7.4% (7/94) and 5.6% (15/270), respectively, which were both significantly higher (p < 0.05) than that in the SR/ IR group (2.5%, 38/1539). All DILI cases presented with a hepatocellular pattern. Compared with non-DILI patients, DILI subjects had longer hospital stays and greater hospital expenses in the SR/IR population (both p < 0.05) ( Table 2).

Risk Factors of DILI in SR/IR Patients.
e demographic and laboratory characteristics of SR/IR patients and univariate analysis are shown in Table 3. e occurrences of DILI in SR/IR patients were primarily associated with age (≤5.2 years), treatment course (≥5), and baseline serum parameters before treatment (cystatin C > 0.79 mg/L, albumin ≤45 g/L, and GGT >17 U/L) ( Table 4). e treatment course and baseline GGT were assigned a value of 2, and the remaining factors were denoted as 1 if they reached the above threshold values, otherwise they were assigned as 0. e area under the ROC curve was 0.846 (95% CI 0.827-0.863) with a cutoff value of 3, and the sensitivity and specificity were 94.7% and 62.3%, respectively ( Figure 2).    Canadian Journal of Gastroenterology and Hepatology

Risk Factors of DILI in HR Patients.
e descriptive statistics of HR patients and univariate analysis are shown in Table 5. e baseline albumin levels were significantly lower (p < 0.05) in the DILI group than in the non-DILI group, whereas GGTand ASTwere significantly higher (both p < 0.05).

Drug Causality Assessment Using the Updated RUCAM.
As all cases with liver injury presented with a hepatocellular type, the RUCAM worksheet for hepatocellular injury was undertaken to assess the strength of association between drug exposure and liver injury [6]. For each patient with suspected DILI, based on the defined criteria, the score of each item was assigned, the sum of which provided a final score with causality grading that reflects the likelihood that the hepatic injury is due to a specific medication: ≥9, highly probable; 6∼8, probable; 3∼5, possible; 1∼2, unlikely; and ≤0, excluded.

Discussion
e cause of DILI is not completely understood but potentially involves multiple host demographic, clinical, and laboratory features [7]. However, the intricate interplay between these features may yield different conclusions concerning DILI risk factors in diverse research contexts [8]. As for childhood ALL patients, the clinical conditions of the patients, existing comorbidities, and polypharmacy make the assessment particularly challenging. In this study, we evaluated these factors at the DILI onset during consolidation therapy in childhood. HD-MTX treatment and preexisting liver conditions were considered as common underlying risk factors in patients receiving SR, IR, HR-1, and HR-2 regimens.
Indeed, it is generally considered that elevations of serum aminotransferases after HD-MTX are transient and reversible and do not lead to chronicity of liver disease [9]. However, we found that DILI subjects in the SR/IR group had longer hospital stays and greater hospital expenses than the non-DILI patients, although no significant differences were observed in HR subgroups, indicating that the development of DILI may increase the time cost and economic burden of leukemia treatment. is shows that it is warranted to study DILI during consolidation therapy in childhood ALL. e overall incidence of DILI in HR patients (5.6%, 15/ 270) was significantly higher (p < 0.05) than that in SR/IR patients (2.5%, 38/1539), which was paralleled by increased exposure to drugs, implying that the more chemotherapy agents the patients received, the higher incidence of DILI occurred. is result is in line with what has been reported in the literature [10]. On the other hand, the incidence of DILI in IR patients who received 5 g/m 2 of HD-MTX monotherapy was found to be 2.9% (25/871) and was significantly lower (p < 0.05) than that in the HR-1 group (7.4%, 7/94). However, no statistically significant difference (p > 0.05) was observed in the incidence of DILI between HR-2 (5.5%, 5/91) and IR groups (2.9%, 25/871). ese results indicate that, apart from HD-MTX, the concomitant drugs may also play a contributing role in the development of DILI in HR patients.
No. of cases HD-MTX HD-Ara-C PEG-ASP HR-1 7 7 (100.0%) 4 (57.0%) 0 (0.0%) HR-2 5 5 (100.0%) NA 1 (20.0%) HR-3 3 NA 3 (100.0%) 1 (33.3%) updated RUCAM, a validated causality assessment method, is the most commonly used tool worldwide for assessing causality in suspected DILI cases [11]. Concomitant/sequential exposure to multiple drugs is a prominent feature during consolidation therapy for HR patients. erefore, it would be essential to identify which drug is most likely to cause liver injury in this category of patients. In this study, we performed separate causality assessment for each implicated drug with the updated RUCAM. To assess the strength of association between liver injury and the suspected drugs, we particularly focused on the high causality gradings of "probable" (score 6∼8) or "highly probable" (score ≥9) likelihood scores. Within the DILI cases for evaluation, HD-MTX, HD-Ara-C, and PEG-ASP were identified as having higher RUCAM scores than other concomitant drugs. In general, 100% (12/12) of HD-MTX in "HR-1 + HR-2", 70% (7/10) of HD-Ara-C in "HR-1 + HR-3," and 13% (2/15) of PEG-ASP in "HR-1 + HR-2 + HR-3" were assessed as high-degree causality (probable or highly probable). It can be concluded that HD-MTX, HD-Ara-C, and PEG-ASP are the most likely causative agents of DILI in HR patients. In this study, as HR-1, HR-2, and HR-3 patients underwent combination therapy regimens containing "HD-MTX + HD-Ara-C," "HD-MTX + PEG-ASP," and "HD-Ara-C + PEG-ASP," respectively, this might partially explain why HR patients had a higher incidence of DILI than the SR/ IR subjects who received single HD-MTX monotherapy. It is worth noting that exclusion of alternative causes is a crucial element of causality assessment in a suspected DILI. In this study, by using the updated RUCAM methodology (element: search for alternative causes), we carefully assessed the non-drug-related etiologies (such as alcohol, viral infections, and others) that could lead to a similar pattern of liver injury. All these potential alternative causes were reasonably ruled out for our patients after thorough investigation.
In host-related factors, the most significant factor identified in our study was the presence of raised liver enzymes before chemotherapy, which coincides with previous reports [12]. In this situation, patients' liver function tends to aggravate when challenged with specific chemotherapy drugs. Regarding the age, our results showed that, for SR and IR patients, younger children (≤5.2 years) might have a higher risk of DILI induced by HD-MTX. Previous study has shown that female patients appear to be more susceptible than male patients to DILI following exposure to certain therapeutic drugs, and a number of hypotheses have been put forward to explain the sex bias in susceptibility to DILI, including pharmacokinetics or pharmacodynamics, specific hormonal effects, and differences in the response of the immune system to drugs [13]. However, differences in male and female incidence rates were not observed in SR/IR (23/981 vs. 15/558, p > 0.05) or HR groups (6/171 vs. 9/99, p > 0.05) in this study. Similar to our results, a US Drug-Induced Liver Injury Network (DILIN) prospective study failed to show apparent correlation between female sex and severity of DILI [14].
is discrepancy indicates that the effect of gender on DILI incidence may vary depending on the clinical contexts such as study designs, interventions, patient populations, and regions. In addition, the components of metabolic syndrome (e.g., obesity and diabetes mellitus) are considered risk factors for drug-associated fatty liver disease (DAFLD) in patients treated with MTX [8]. However, in this study, we did not observe significant difference in BMI between the DILI group and the non-DILI group, possibly due to the different clinical conditions in the present study and those reported by other investigators [15,16].
Patients of SR/IR receiving five or more courses of therapy exhibited a higher risk of DILI, which highlights that more attention should be paid as the dosing frequency and cumulative dose of MTX increase. Our data also showed that baseline cystatin C had a positive association with DILI, which has rarely been reported in previous studies.
Previous studies have shown that adjuvants like folic acid and vitamin E can reduce the incidence of MTX hepatotoxicity in rheumatoid arthritis [17][18][19]. However, whether hepatoprotective drugs (e.g., reduced glutathione, glycyrrhizin, and glucurolactone) reduce the occurrence of DILI during consolidation chemotherapy for childhood ALL remains largely unclear. At present, there are no available data concerning their effects on the efficacy of chemotherapy, particularly for regimens containing HD-MTX. In addition, the potential interaction between hepatoprotective drugs and chemotherapy agents may increase the complexity of metabolism of the latter. erefore, whether the prophylactic use of hepatoprotective drugs should be recommended and routinely administered before chemotherapy in ALL clinical practice remains questionable. In our study, patients receiving these hepatoprotective drugs did not exhibit reduced DILI, indicating that such agents may not be required before consolidation therapy.
is study has a few limitations that need to be considered. First, due to the retrospective nature of this study, missing data are a well-known source of bias in research. For example, in this study, two cases of DILI in the SR/IR group were assessed as "possible" causality because their biochemical tests were not performed at certain time points after the treatment, which would decrease RUCAM scores thereby leading to underestimation in the causality assessment. Second, it should be noted that the role of other potential risk factors cannot be excluded. As is known, inheritance or genetic variations are crucial contributors to DILI [20,21]. For instance, human leukocyte antigen (HLA) alleles have been suggested as strong risk factors for the development of DILI with a range of drugs [22]. Methylenetetrahydrofolate reductase (MTHFR) C677T gene polymorphisms have also been associated with hepatotoxicity due to MTX [8]. However, the effect of such genetic markers on DILI was not addressed in this study. ird, limited by the resources currently available, the number of courses recruited for HR patients was relatively limited. Further studies with larger case numbers need to be conducted for a better understanding of DILI in HR patients. Fourth, apart from liver biochemical blood tests, recent evidence has highlighted the importance of ultrasound evaluation and measurement of hepatic stiffness in ALL patients treated with hepatotoxic drugs. ese tools are important to support Canadian Journal of Gastroenterology and Hepatology 7 screening of other forms of liver-related adverse effects, including DAFLD in patients treated with MTX and portal hypertension-related complications (e.g., sinusoidal obstruction syndrome/veno-occlusive disease (SOS/VOD) and hepatomegaly) in patients treated with inotuzumab ozogamicin [8,23]. e lack of these noninvasive assessments of the liver is another possible limitation of our study and needs to be studied in future work.
In conclusion, the occurrences of DILI in SR and IR patients, who received HD-MTX monotherapy, were primarily associated with age, treatment course, and baseline serum parameters before treatment (e.g., cystatin C, albumin, and GGT). As for HR patients, preexisting liver conditions prior to treatment were potential risk factors of DILI. HD-MTX, HD-Ara-C, and PEG-ASP were the most likely causative agents of DILI during consolidation therapy.

Data Availability
e data used to support the findings of this study are included within the article.

Ethical Approval
is research was ethically approved by the Ethical Review Committee of the First Affiliated Hospital of Guangxi Medical University.

Conflicts of Interest
e authors declare that there are no conflicts of interest.

Authors' Contributions
Fu-Li Qin, Guo-You Sang and Xiao-Qin Zou are contributed equally to this work.