Alcohol-related cancer morbidity and mortality are stratified using modified albumin platelet product

Alcohol abuse is associated with several diseases, such as hepatocellular carcinoma, cirrhosis, and extrahepatic malignancies. Recently, we reported albumin platelet product (APP) and modified APP (mAPP) as novel indices of liver fibrosis staging and prognosis in patients without alcoholic liver diseases. This retrospective cohort study aimed to extend application of APP and mAPP in prognosis prediction of patients with alcoholic liver diseases. We enrolled 222 patients with alcoholic liver diseases based on their medical records. Cut-off values of APP = 4.349 and mAPP = 2.484 were adopted based on a past report. Hazard ratios of APP and mAPP were compared to those of albumin-bilirubin score and fibrosis-4 index. The primary and secondary endpoints were carcinogenesis and death, respectively. Thus, APP = 4.349 and mAPP = 2.484 significantly differentiated cancer-free survival and overall survival in univariate analysis. Hazard ratios of mAPP = 2.484 were greater than those of the albumin-bilirubin score of -2.270 and fibrosis-4 index of 3.25. Multivariate analysis revealed mAPP = 2.484 as an independent risk factor for carcinogenesis and overall death. In conclusion, mAPP is a simple index to stratify patient’s risk for carcinogenesis and death.


Baseline characteristics of patients
The median age of the patients was 60 years, predominantly comprising men (Table 1).The median gammaglutamyl transferase (GGTP) values were above the upper limit of the normal range for both male and female patients.The median platelet count was above 150 × 10 9 /L.The median APP and mAPP values exceeded their cut-off values for cirrhosis (above 4.349 and 2.484 for APP and mAPP, respectively).

Patient follow-up
The patients were followed-up for a median of 1492 days, with a first quartile value of 615 days (Table 2).A total of 41 patients were diagnosed with complications of malignancies, 23 patients were complicated with HCC.Colon, gastric, prostate, and oropharyngeal cancers were diagnosed in four, three, three, and two patients, respectively.Moreover, cholangiocellular, esophageal, duodenal, lung, and thyroid cancer and parotid tumor were diagnosed

Univariate analysis for prognosis prediction
A univariate analysis was conducted to determine the prognostic abilities of APP and mAPP, compared to that of the albumin-bilirubin (ALBI) score and fibrosis-4 (FIB-4) index.Both APP = 4.349 and mAPP = 2.484 significantly stratified the cancer-free survival prognosis of patients, with hazard ratios of 2.207 (P = 0.0031) and 2.627 (P = 0.0002) (Fig. 1).Compared to that of APP, mAPP had a better hazard ratio.The ALBI score of − 2.270 and FIB-4 index of 3.25 were able to significantly stratify the survival curves with lower hazard ratios (1.923 for ALBI = − 2.270, P = 0.0171; 2.181 for FIB-4 = 3.25, P = 0.0046) than that of mAPP (Supplementary Figs.2a, 4a).
In addition, predictive values of HCC-free survival were compared among APP, mAPP, ALBI score, and FIB-4.As shown in Fig. 2a,b, APP and mAPP significantly stratified the cohort into two groups, a group with better HCC-free survival and the worse one.The hazard ratios were 2.184 for APP (P = 0.0111) and 2.563 for Figure 1.Cancer-free survival evaluated using APP and mAPP.A univariate analysis was performed to evaluate the prognostic potential of APP and modified APP (mAPP) for cancer-free survival.APP = 4.349 and mAPP = 2.484 were adopted as cut-off values.(a) APP = 4.349 significantly differentiates the cancer-free survival of patients with alcoholic liver diseases with an HR of 2.207 (P = 0.0031).(b) mAPP was able to predict the prognosis of the patients with a better HR of 2.627 than that of APP (P = 0.0002).HCC, hepatocellular carcinoma; HR, hazard ratio; mAPP, modified albumin platelet product.
Consequently, a mAPP score of 2.484 was the best among APP, mAPP, ALBI grade, and FIB-4 index for stratifying cancer-free, overall, and HCC-free survival rates.Therefore, we focused on mAPP for further evaluation.

Multivariate analysis for prognosis prediction
A multivariate analysis was conducted to determine whether mAPP was an independent risk factor for both carcinogenesis and deaths after adjusting for age, sex, and hemoglobin A1c (HbA1c) (%) using a Cox proportional hazard model.As shown in Table 3, mAPP was an independent risk factor for all endpoints, including carcinogenesis (HR, 2.477; p = 0.0072), carcinogenesis or death (HR, 3.244; p = 0.0003), and overall death (HR, 3.622; p = 0.0007).Age was significantly associated with an increased risk of carcinogenesis (HR, 1.058; p = 0.0003), and of carcinogenesis or death (HR, 1.046; p = 0.0005).However, age was not a significant variable for predicting overall mortality (p > 0.05).
HbA1c > 6.2% also acted as an independent risk factor for the outcomes when an endpoint was put at carcinogenesis (HR, 2.352; p = 0.0236), or at carcinogenesis or death (HR, 2.283; p = 0.0147).HbA1c was not included in a multivariate analysis for overall death according to a guideline 6 .A similar analysis for HCC complication was omitted for the same rationale 6 .Harrell's C-index was approximately 0.7 for all calculations.

GGTP reduction rate from baseline to outcome
To estimate whether habitual alteration of alcohol consumption correlated with patient outcomes, the GGTP reduction rate from baseline to outcome was calculated.In 41 patients with carcinogenesis during the observation period, the GGTP values at diagnosis of malignancies were compared to those at baseline.In 167 patients whose observation was censored alive with no malignancies during the observation period, GGTP at censor was compared to that at baseline.As shown in Supplementary Fig. 6, the GGTP levels were more reduced in patients without malignancies compared to those of patients complicated with malignancies, as a trend (p > 0.05).

Discussion
The current study was conducted to investigate the potential of APP and mAPP for managing patients with alcoholic liver diseases.APP and mAPP were established based on an analysis of patients with non-alcoholic etiologies.The present study revealed that (1) APP and mAPP can stratify cancer-free survival and overall survival, with mAPP showing better performance, and that (2) mAPP is an independent risk factor for carcinogenesis and deaths.
Alcohol is considered the most severe carcinogen by the International Agency for Research on Cancer since 1988 (https:// publi catio ns.iarc.fr/ Book-And-Report-Series/ Iarc-Monog raphs-On-The-Ident ifica tion-Of-Carci nogen ic-Hazar ds-To-Humans/ Alcoh ol-Drink ing-1988).Among newly diagnosed cancers worldwide, an estimated 4.1% of cancers (741,300 cases) in 2020 were attributed to alcohol consumption 7 .Alcoholic beverages consumption is related to cancers in specific organs, including the oral cavity, pharynx, larynx, esophagus, and liver.The prevalence of colorectal cancer, female breast cancer, and pancreatic cancer also increase in accordance with alcohol intake.
Alcohol drives carcinogenesis via several pathways, including acetaldehyde production, excess oxidative stress, alteration of gut microbiome, and dysregulated retinoid metabolism 8 .Acetaldehyde, a metabolites of ethanol, has affinity to DNA.Acetaldehyde directly binds to DNA and may change DNA synthesis and repair 9 .Acetaldehyde results in DNA methylation through inhibition of DNA methyltransferase 9 .The process of oxidizing ethanol to acetaldehyde causes oxidative stress, which provides the basis of carcinogenesis 10 .Ethanol metabolism by CYP2E1 produces excess reactive oxygen species known as superoxide anion and hydrogen peroxide, which contributes to liver, colorectal, and esophageal carcinogenesis 11,12 .Liver cancer will be initiated by habitual alcohol consumption partially via altered gut microbiome in the context of gut-liver axis 13 .Chronic alcohol intake reduces the blood retinol levels, which enhances carcinogenesis in the head and neck regions 14 .
APP and mAPP were initially established using a training cohort of patients with a biopsy-proven staging of fibrosis 5 , with a cut-off value for cirrhosis determined as 4.349 and 2.484 for APP and mAPP, respectively.The cut-off value of APP stratified HCC-free survival and overall survival.A validation cohort confirmed that APP was able to diagnose cirrhosis and predict the prognosis of patients.However, the training and validation cohorts comprised patients with non-alcoholic liver diseases.Therefore, we aimed to extend the application of APP and mAPP to patients with alcoholic liver diseases.
Alcoholic liver diseases have a spectrum from simple steatosis to hepatitis and cirrhosis 15 .The current cohort may have presented with mild alcoholic changes in the liver based on their baseline platelet count and chemistry panel 16 .As described in a clinical guideline as an early alcoholic liver disease, patients in the current study were characterized with minimal elevation of AST and ALT, elevated GGTP with normal T-Bil levels, and prothrombin time (PT)-international normalized ratio (INR) 17 .Most patients had a model for end-stage liver disease (MELD) score of < 10 points at baseline (Table 2).Thus, evaluation of baseline liver function using Maddrey's discrimination function was omitted in the current study 18,19 .
However, a high proportion of patients with histologically confirmed alcoholic liver diseases may remain asymptomatic and show no laboratory abnormalities 20 .In the current cohort, approximately 10% of the patients died from cirrhosis or HCC within the observational period, as shown in Table 2.The around 10% of liver-related mortality in the current observation was orchestrated with a past research 1 .Our findings suggest that mAPP can effectively stratify the prognosis of patients, even those in the early stages of alcoholic liver diseases.
In a multivariate analysis with endpoints of carcinogenesis, and carcinogenesis or death, HbA1c > 6.2% was extracted as an independent risk factor accompanied with mAPP > 2.484 (Table 3).Diabetes mellitus is an established risk factor for cancer prevalence in several organs 21 .According to a past report, alcohol abusers tend to be complicated with diabetes mellitus 22 .Habitual drinking will clinically cause chronic pancreatitis, resulting in impaired insulin secretion from pancreatic beta cells 23 .mAPP has several strengths, including the incorporation of serum albumin in its equation.Serum albumin exhibits a stronger association with the prognosis of patients with alcoholic liver diseases compared to the association exhibited by pathological findings in liver tissues 24 .The second strength of mAPP is that it considers platelet count in its calculation.As the decrease in platelet count correlates with liver fibrosis progression and portal hypertension, mAPP can detect liver fibrosis more sensitively than ALBI grade can 25 .Furthermore, alcohol may induce thrombocytopenia by transiently regulating the spleen's store of platelets 26 .The third strength lies its simple equation excluding natural logarithm or square root, compared to the ALBI score and FIB-4 index.
A limitation of the current study might be that patients were selected from a tertiary hospital, where a relatively limited number of alcoholic patients were referred from clinics and hospitals in the community.Therefore, the clinical application of mAPP in patients with alcoholic liver diseases should be validated using a more general population of alcohol abusers.In addition, the cut-off values of APP, mAPP, ALBI score, and FIB-4 index were derived from the cut-off values got cirrhosis of different etiologies: APP = 4.349 and mAPP = 2.484 pertain to a mainly HCV infection, an ALBI score = − 2.270 pertains to HCV infection, and an FIB-4 index = 3.25 pertains to non-alcoholic steatohepatitis.Optimal cut-off values for prognosis, not for cirrhosis, should be further searched in patients with alcoholic liver diseases based on a larger cohort.Finally, accumulated doses of alcohol www.nature.com/scientificreports/consumption over time were not incorporated into the study.Alteration of the GGTP levels between the baseline and outcome was considered as a surrogate for it.
In conclusion, mAPP can predict cancer-free and overall survival.The simple index can serve as a convenient measure for managing patients with alcohol abuse.

Ethics
This study was conducted in accordance with the ethical principles of the Declaration of Helsinki and was approved by the Institutional Review Board of Kagawa University, Faculty of Medicine (Heisei-30-151) 27 .Informed consent was obtained from all patients for the analysis of their clinical data.In cases where patients had died and had no listed relatives in their medical records, we provided opt-out methods for the relatives of the dead participants by publishing a summary of this study on our university website 28,29 .

Study design
The current study was a retrospective observational study conducted at an outpatient clinic of internal hepatology at a single university hospital.The baseline of observation was established at the time of the diagnosis of alcoholic liver disease.The primary endpoint was the occurrence of malignancies in any organ, except for carcinomas in situ of the gastrointestinal tract.The secondary endpoint was defined as death from any cause.Cut-off values for APP and mAPP were determined based on the diagnostic thresholds for cirrhosis in patients with non-alcoholic etiologies, as reported in a previous study (APP = 4.349 and mAPP = 2.484 for cirrhosis) 5 .
To compare the diagnostic abilities of the two indices with the ALBI grade, we adopted a cut-off value between ALBI grades 2a and 2b (ALBI score = − 2.270) 30 .An ALBI score of − 2.270 was an approximation of a cut-off ALBI score = − 2.125 for cirrhosis in patients with hepatitis C infection 31 .For FIB-4, a cut-off value of 3.25 was employed as an indicator of high risk of cirrhosis, as reported in a past study 32 .
HbA1c of 6.2% was determined as a threshold of impaired glucose tolerance with or without diabetes treatment.In patients with no record of HbA1c, their HbA1c values were assumed to be < 6.2%.The study was conducted in accordance with the STROBE statement 33 .

Patients
Patients were recruited based on a medical record.Patients who were diagnosed as alcoholic liver diseases, including alcoholic steatosis, alcoholic hepatitis, or alcoholic cirrhosis, between January 1, 2002 and December 31, 2021 were registered.After reviewing their clinical data, patients with alcoholism were included in the study, while those with any confounding factors at presentation were excluded.Specifically, patients with hepatitis viral infection, autoimmune hepatitis, or other specific etiologies were excluded.Patients with HCV were defined as those with a history of positive HCV-RNA or anti-viral therapy.Patients with spontaneous remission of HCV, defined as those with positive anti-HCV antibodies, negative HCV-RNA, or no history of viral eradication therapy, were included in the current cohort.Patients with HBV infection were defined as those positive for HBV surface (HBs) antigen and HBV-DNA, while those with functional cure without any history of anti-viral therapy were included.The functional cure was defined as a negative status for HBs antigen or HBV-DNA and a positive status for HBV core (HBc) and HBs antibody.Patients with autoimmune hepatitis were identified based on the criteria of International Autoimmune Hepatitis Group scores ≥ 10.
Patients with malignancies at the diagnosis of alcoholic liver diseases were excluded, whereas malignancies were defined as all malignant diseases except for carcinomas in situ of the gastrointestinal tract.Patients who were diagnosed with malignancies within 30 days after the diagnosis of alcoholic liver diseases were also excluded.

Clinical data
The following clinical data were extracted from the patient's medical records: age, sex, history of alcohol consumption, and medications.For laboratory data, platelet count, aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, GGTP, T-Bil, HbA1c (%), and PT-INR.T-Bil (mg/dL) was converted to T-Bil (µmol/L) according to the following equation: T-Bil (mg/dL) × 17.2.HbA1c (%) as described by Japan Diabetes Society (JDS) was converted to values based on the National Glycohemoglobin Standardization Program (NGSP) 34 .

Statistical analyses
Continuous variables were presented as medians and interquartile ranges and analyzed using the Mann-Whitney U test or Spearman's rank correlation coefficient.The statistical analyses above were performed using GraphPad Prism 6 (GraphPad Software, La Jolla, CA).A Cox proportional hazard model was analyzed using EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan), a graphical user interface for R software (The R Foundation for Statistical Computing, Vienna, Austria) 38,39 .The number of variables in a multivariate analysis was controlled following a rule that number of events should be as 10 times as number of variables in proportional hazard ratio model 6 .P < 0.05 was considered statistically significant.

Table 1 .
Baseline characteristics of patients.APP, Albumin platelet product; ALBI score, albumin bilirubin score; IQR, interquartile range; MELD score, model for end-stage liver disease score.

IQR) or case number
in one patient each.The median time from baseline to carcinogenesis was 1819 days.A total of 29 patients died during the follow-up period.Among them, 12 had cirrhosis, and eight had HCC due to alcoholic cirrhosis.The median time from baseline to death was 1937 days.