Prognostic significance of nonprotein respiratory quotient in patients with liver cirrhosis.

The aim of this study was to examine the effect of nonprotein respiratory quotient (npRQ), as assessed using indirect calorimetry, on clinical outcomes in patients with liver cirrhosis (LC). A total of 244 LC patients were evaluated in this study. For the univariate analysis, for each continuous variable, the optimal cutoff value that maximized the sum of sensitivity and specificity was selected using receiver operating curve (ROC) analysis for survival. There were 137 men and 107 women with the median (range) age of 67 (25-90) years. Indirect calorimetry indicated that 54 patients (22.1%) had hepatocellular carcinoma (HCC) on radiological findings and 59 patients (24.2%) had protein energy malnutrition, as defined by npRQ <0.85 and serum albumin level <3.5 g/dL. In ROC analysis of npRQ for survival, the optimal cutoff point of npRQ was 0.849 for all cases (area under the ROC = 0.61272; sensitivity, 66.22%; and specificity, 57.06%). The median follow-up periods after indirect calorimetry were 4.35 years (range, 1.01-9.66 years) in patients with npRQ ≥0.85 (n = 122) and 3.71 years (range, 0.19-9.51 years) in patients with npRQ <0.85 (n = 122). The 1-, 3-, and 5-year cumulative OS rates in patients with npRQ ≥0.85 were 100%, 87.79%, and 77.24%, respectively, whereas those in patients with npRQ <0.85 were 94.26%, 73.65% and 57.78%, respectively (P = 0.0004). In the multivariate analysis, presence of HCC (P = 0.0045), body mass index (P < 0.0001), serum albumin (P = 0.0441), prothrombin time (P = 0.0463), npRQ (P = 0.0024), estimated glomerular filtration rate (P = 0.0086), and des-γ-carboxy prothrombin (P = 0.0268) were found to be significant predictors associated with OS. For all cases, risk stratification for survival was well performed using these significant variables. In conclusion, npRQ value, as assessed by indirect calorimetry, can be helpful for predicting clinical outcomes for LC patients.

Introduction

The liver is an essential organ for the metabolism of 3 major classes of molecules: fat, protein, and carbohydrate.[ Liver cirrhosis (LC) is known to be a terminal form of chronic liver disease and is accompanied by numerous nutritional disorders.[ In addition, LC is often complicated with protein-energy malnutrition (PEM), which is associated with clinical outcomes in LC patients.[ Energy malnutrition can be assessed by measuring the nonprotein respiratory quotient (npRQ) using indirect calorimetry.[ RQs reflect which macronutrients are being metabolized: values that approach 0.7 indicate that lipids are being consumed and values that approach 1.00 indicate that carbohydrates are largely being burned.[ A previous study reported that npRQ correlated significantly with arm circumference and arm muscle circumference but not with triceps skinfold thickness.[ Another study demonstrated that the plasma level of free fatty acid was closely correlated with npRQ value.[

Conversely, various nutritional therapies for LC have been recently proposed and validated. These include branched-chain amino acid (BCAA) administration therapy, late evening snack (LES) with BCAA enriched snack, carnitine therapy, zinc replacement therapy, oral diet or exercise.[ Particularly, BCAA therapy and LES have originated from Japan and many Japanese researchers have reported their usefulness in the clinic.[ In addition, in hepatitis virus related LC patients, antiviral therapies have been recommended in recent years.[

In 2002, Tajika et al reported that the overall survival (OS) rate in patients with LC was significantly lower in subjects with npRQ <0.85 than in subjects with npRQ ≥0.85 (P < 0.01) and that npRQ was a useful predictor for LC patients.[ However, as previously mentioned, over the period of more than 10 years, nutritional interventional therapies for LC have dramatically changed and thus nutritional status in LC patients has improved.[ In the era of novel nutritional therapies for LC, whether npRQ can be a useful predictor for patients with LC remains unknown.[

The aim of this study was therefore to examine the impact of npRQ on clinical outcomes in patients with LC.

Patients and methods

Patients

Between August 2006 and January 2015, a total of 300 LC individuals with available data for indirect calorimetry were admitted at the Division of Hepatobiliary and Pancreatic disease, Department of Internal Medicine, Hyogo College of Medicine, Hyogo, Japan. Of these, 56 patients had been lost to follow up within 1 year after performing indirect calorimetry and were excluded from the analysis. Thus, a total of 244 LC patients were evaluated in this study. Follow-up observation after indirect calorimetry consisted of regular blood examinations and radiological evaluation by ultrasonography, computed tomography, or magnetic resonance imaging to detect hepatocellular carcinoma (HCC) every 3 to 6 months. LC was diagnosed radiologically and/or pathologically. In patients with serum albumin level <3.5 g/dL, BCAA granule therapy or LES with BCAA enriched snacks were considered.[ In hepatitis virus related LC patients, antiviral therapies were considered.[ We retrospectively examined baseline npRQ levels and other clinical variables on OS and examined variables associated with OS in the univariate and multivariate analyses. The npRQ level using indirect calorimetry was calculated, as reported previously.[

The present study comprised a retrospective analysis of patient clinical data. The ethics committee in our hospital approved the present study protocol and this study protocol strictly adhered to all provisions of the Declaration of Helsinki.

HCC diagnosis and HCC therapy

HCC was diagnosed radiologically or histologically based on the European Association for the Study of Liver guideline.[ In some subjects who presented with atypical liver nodules, we conducted ultrasonography-guided tumor biopsy. As for HCC therapy, the most appropriate HCC therapy for each case, such as surgical resection, locoregional therapies, and systemic chemotherapy, including sorafenib, was chosen through a discussion with hepatologists, radiologists, and surgeons. In cases of HCC recurrence, the same strategy was chosen in each patient.

Statistical analyses

Categorical parameters were compared by Fisher exact test. Continuous parameters were compared by unpaired t test or Mann–Whitney U test as appropriate. In continuous variables, receiver operating characteristic curve (ROC) analysis for survival was performed by selecting the optimal cutoff point that maximized the sum of sensitivity and specificity. The cutoff points for each subject were used to divide the study population into 2 groups, which were then treated as dichotomous covariates in the univariate analysis. OS curves were generated using the Kaplan–Meier method and compared using the log-rank test. Parameters with a P-value <0.05 in the univariate analysis were entered into the multivariate analysis in the Cox proportional hazards model. OS was defined as the time interval from the date of performing indirect calorimetry until death from any cause or the last follow-up. Data are expressed as the median value (range). Values with P < 0.05 were regarded as those with statistical significance. Statistical analysis was performed with the JMP 11 (SAS Institute, Inc., Cary, NC).

Results

Baseline characteristics

The baseline characteristics of the evaluated subjects (n = 244) are presented in Table 1. There were 137 males and 107 females with a median (range) age of 67 (25–90) years. In terms of causes of LC, hepatitis B virus-related LC was found in 20 patients, hepatitis C virus-related LC in 143 patients, alcoholic liver disease-related LC in 29 patients and other causes in 52 patients. There were 152 patients with a Child–Pugh A score, 86 with a Child–Pugh B score and 6 with a Child–Pugh C score. The model for end-stage liver disease score ranged from −5.0 to 23.4 (median, 5.0). In this study, the npRQ ranged from 0.663 to 1.677 (median: 0.850). While performing indirect calorimetry, 54 patients (22.1%) were observed to have HCC on radiological findings. ROC analyses of npRQ for survival indicated that the optimal cutoff points of the npRQ score was 0.849 for both all cases (area under the ROC (AUROC) = 0.61272; sensitivity, 66.22%; and specificity, 57.06%) and for patients without HCC at indirect calorimetry (n = 190) (AUROC = 0.65678; sensitivity, 76.74%; and specificity, 57.14%) (Fig. 1). In the present study, 59 out of 244 patients (24.2%) had PEM as defined by npRQ <0.85 and serum albumin level <3.5 g/dL.[

Table 1

Baseline characteristics (n = 244).

Figure 1

Receiver operating curve (ROC) analyses of npRQ for survival. (A) The optimal cutoff point of npRQ was 0.849 for all cases (area under the ROC (AUROC) = 0.61272; sensitivity, 66.22%; and specificity, 57.06%). (B) The optimal cutoff point of npRQ for patients without HCC, as indicated by indirect calorimetry (n = 190), was 0.849 (AUROC = 0.65678; sensitivity, 76.74%; and specificity, 57.14%).

Comparison of OS in patients with npRQ ≥0.85 and npRQ <0.85 for all cases (n = 244)

The median follow-up periods after indirect calorimetry were 4.35 years (range, 1.01–9.66 years) in patients with npRQ ≥0.85 (n = 122) and 3.71 years (range, 0.19–9.51 years) in patients with npRQ <0.85 (n = 122). The 1-, 3-, and 5-year cumulative OS rates in patients with npRQ ≥0.85 were 100%, 87.79%, and 77.24%, respectively, while those in patients with npRQ <0.85 were 94.26%, 73.65%, and 57.78%, respectively (P = 0.0004) (Fig. 2).

Figure 2

Cumulative overall survival for all cases based on npRQ value. The 1-, 3-, and 5-year cumulative overall survival rates in patients with npRQ ≥0.85 (n = 122) were 100%, 87.79%, and 77.24%, respectively, while those in patients with npRQ <0.85 (n = 122) were 94.26%, 73.65%, and 57.78%, respectively (P = 0.0004).

Comparison of OS in patients with npRQ ≥0.85 and npRQ <0.85 for patients without HCC, as indicated by indirect calorimetry (n = 190)

We also performed subanalyses in patients excluding patients that have HCC, as indicated by indirect calorimetry (n = 190). The median follow-up periods after indirect calorimetry were 4.47 years (range, 1.17–9.66 years) in patients with npRQ ≥0.85 (n = 94) and 3.75 years (range, 0.19–9.51 years) in patients with npRQ <0.85 (n = 96). The 1-, 3-, and 5-year cumulative OS rates in patients with npRQ ≥0.85 were 100%, 95.49%, and 87.98%, respectively, while those in patients with npRQ <0.85 were 93.75%, 80.84%, and 62.5%, respectively (P < 0.0001) (Fig. 3).

Figure 3

Cumulative overall survival for patients without HCC, as indicated by indirect calorimetry, based on npRQ value. The 1-, 3-, and 5-year cumulative overall survival rates in patients with npRQ ≥0.85 (n = 94) were 100%, 95.49%, and 87.98%, respectively, while those in patients with npRQ <0.85 (n = 96) were 93.75%, 80.84%, and 62.5%, respectively (P < 0.0001).

Causes of death

Seventy-four patients (30.3%) died during the follow-up period. The causes of death were liver failure in 29 patients, HCC progression in 37 patients and miscellaneous causes in 8 patients.

Univariate and multivariate analyses of factors contributing to OS for the entire cohort

Univariate analysis identified the following factors as significantly associated with OS for the entire cohort (n = 244): age ≥70 years (P = 0.0008); presence of HCC (P < 0.0001); Child–Pugh A (P = 0.0091); aspartate aminotransferase (AST) ≥56 IU/L (P = 0.0359); alanine aminotransferase ≥59 IU/L (P = 0.0063); serum albumin ≥3.7 g/dL (P = 0.0046); prothrombin time (PT) ≥77.3% (P = 0.0036); triglyceride ≥72 mg/dL (P = 0.0028); estimated glomerular filtration rate (eGFR) ≥78.35 mL/min/1.73 m2 (P = 0.0028); body mass index (BMI) ≥22.2 kg/m2 (P = 0.0001); npRQ ≥0.85 (P = 0.0004); alpha-fetoprotein (AFP) ≥5.7 ng/mL (P = 0.0002); and des-γ-carboxy prothrombin (DCP) ≥40 mAU/mL (P < 0.0001) (Table 2). The hazard ratios (HRs) and 95% confidence intervals (CIs) calculated by using multivariate analysis for the 13 significant variables (P < 0.05) in the univariate analysis are presented in Table 3. BMI (P < 0.0001), npRQ (P = 0.0024), presence of HCC (P = 0.0045), serum albumin (P = 0.0441), PT (P = 0.0463), eGFR (P = 0.0086), and DCP (P = 0.0268) were revealed to be significant predictors related to OS in multivariate analysis (Table 3).

Table 2

Univariate analyses of factors linked to overall survival for all cases (n = 244).

Table 3

Multivariate analyses of factors linked to overall survival for all cases (n = 244).

Univariate and multivariate analyses of parameters contributing to OS in patients without HCC, as indicated by indirect calorimetry (n = 190)

Univariate analysis identified the following factors as significantly linked to OS for patients without HCC, as indicated by indirect calorimetry (n = 190): age ≥71 years (P = 0.0001); Child–Pugh A (P = 0.0039); AST ≥49 IU/L (P = 0.0078); serum albumin ≥3.7 g/dL (P = 0.0162); PT ≥77.3% (P = 0.0033); triglyceride ≥68 mg/dL (P = 0.0035); BCAA to tyrosine ratio (BTR) ≥3.55 (P = 0.0257); eGFR ≥85.67 mL/min/1.73 m2 (P = 0.0270); BMI ≥22.3 kg/m2 (P = 0.0002); npRQ ≥0.85 (P < 0.0001); and DCP ≥26 mAU/mL (P = 0.0003) (Table 4). The HRs and 95% CIs calculated by using the multivariate analysis for the eleven significant predictors (P < 0.05) in the univariate analysis are shown in Table 5. BMI (P = 0.0004), npRQ (P = 0.0053), serum albumin (P = 0.0022), and ALT (P = 0.0018) were found to be significant predictors associated with OS in the multivariate analysis (Table 5).

Table 4

Univariate analyses of factors linked to overall survival in patients without hepatocellular carcinoma at indirect calorimetry (n = 190).

Table 5

Univariate and multivariate analyses of factors linked to overall survival in patients without hepatocellular carcinoma at baseline (n = 190).

Comparison of baseline characteristics in patients with npRQ ≥0.85 and npRQ <0.85 for all cases

We also compared baseline characteristics in patients with npRQ ≥0.85 and npRQ <0.85 for all cases. The serum albumin level (P = 0.0049) and PT (P = 0.0166) in patients with npRQ ≥0.85 were significantly higher than those in patients with npRQ <0.85 (Table 6). In other variables, no significant differences were found between the 2 groups.

Table 6

Comparison of baseline characteristics in patients with npRQ ≥0.85 and npRQ <0.85 for all cases (n = 244).

Construction of predictive model for all cases and for patients without HCC, as indicated by indirect calorimetry

Based on significant variables in multivariate analyses, we constructed a predictive model for all cases and for patients without HCC, as indicated by indirect calorimetry. For all cases, patients without HCC, patients with npRQ ≥0.85, patients with eGFR ≥78.35 mL/min/1.73 m2, patients with BMI ≥22.2 kg/m2, patients with serum albumin ≥3.7 g/dL, patients with PT ≥77.3% or patients with DCP <40 mAU/mL were given by 1 point. Total points were calculated in each case. Total points ranged from 0 to 7. Subjects were divided into 3 groups based on total points (5 or more points: Group A (n = 91); 3 or 4 points: Group B (n = 109); and 2 or less point: Group C (n = 44)). Significant differences were found between each of the 2 groups in terms of OS (Group A vs B: P < 0.0001, Group B vs C: P < 0.0001 and Group A vs C: P < 0.0001) (Fig. 4A).

Figure 4

Our proposed predictive model for survival based on significant variables, as determined by multivariate analyses. (A) For all cases. Patients without HCC, patients with npRQ ≥0.85, patients with eGFR ≥78.35 mL/min/1.73 m2, patients with BMI ≥22.2 kg/m2, patients with serum albumin ≥3.7 g/dL, patients with PT ≥77.3% or patients with DCP <40 mAU/mL were given by one point. Total points were calculated in each case. Total points ranged from 0 to 7. Subjects were divided into three groups based on total points (5 or more points: Group A (n = 91); 3 or 4 points: Group B (n = 109); and 2 or less point: Group C (n = 44)). There were significant differences between each of the 2 groups in terms of OS (Group A vs B: P < 0.0001, Group B vs C: P < 0.0001 and Group A vs C: P < 0.0001). (B) For cases without HCC, as indicated by indirect calorimetry. Patients with npRQ ≥0.85, patients with BMI ≥22.3 kg/m2, patients with serum albumin ≥3.7 g/dL and patients with ALT <49 IU/L were given by one point and total points were calculated in each case. Total points ranged from 0 to 4. They were divided into 3 groups based on total points (4 points: Group a (n = 19); 2 or 3 points: Group b (n = 130); and 0 or 1 point: Group c (n = 41)). Significant differences were observed between groups in terms of OS (Group a vs b: P = 0.0455, Group b vs c: P < 0.0001 and Group a vs c: P < 0.0001) (B).

Similarly, for patients without HCC, as indicated by indirect calorimetry, patients with npRQ ≥0.85, patients with BMI ≥22.3 kg/m2, patients with serum albumin ≥3.7 g/dL, and patients with ALT <49 IU/L were given by 1 point and total points were calculated in each case. Total points ranged from 0 to 4. They were divided into 3 groups based on total points (4 points: Group a (n = 19); 2 or 3 points: Group b (n = 130); and 0 or 1 point: Group c (n = 41)). Significant differences were observed between groups in terms of OS (Group a vs b: P = 0.0455, Group b vs c: P < 0.0001 and Group a vs c: P < 0.0001) (Fig. 4B).

Cumulative OS rates in hepatitis C virus (HCV)-related LC patients based on npRQ value

In 143 HCV-related LC patients, 123 patients (86.0%) had a high HCV viral load (≥5 log IU/mL) at baseline. During the follow-up period, 50 patients (35.0%) achieved sustained virological response (SVR). Of these, 31 patients were treated with interferon (IFN)-based therapies, while 19 patients were treated with IFN-free direct acting antiviral (DAA) therapies and only 3 patients died during the follow-up period. The 1-, 3-, and 5-year cumulative OS rates in HCV-related LC patients with SVR were 98.0%, 96.0%, and 93.6%. In terms of OS, the difference in patients who achieved SVR with npRQ ≥0.85 (n = 27) and those with npRQ <0.85 (n = 23) did not reach significance (P = 0.6398).

Discussion

To the best of our knowledge, this is the first validation study for the report of Tajika et al.[ As mentioned previously, in the era of novel nutritional therapies for LC, whether the npRQ can be a useful predictor for patients with LC is unclear.[ Therefore, we conducted this observational study to address this urgent clinical question.

Previously, the proportion for PEM in LC patients was reported to be as high as 65% to 90%.[ While in our present data, 59 out of 244 patients (24.2%) had PEM. These improvements in nutritional status in LC patients may be attributed to the progress for LC therapy over a period of more than 10 years.

In our study, higher npRQ value is significantly related to favorable OS for all cases and for cases without HCC, as indicated by indirect calorimetry, and optimal cutoff points of npRQ for survival in all cases and in patients without HCC, as indicated by indirect calorimetry, are both 0.849, which agree with data in the study by Tajika et al.[ These results suggest that the npRQ value can be helpful for predicting survival in patients with LC and the data in Tajika et al study were validated by our data. Furthermore, our constructed predictive model can be promising for the development of a risk stratification method for predicting survival, although it should be validated in other independent cohorts.

A decrease in the npRQ value indicates a decrease in glycogen storage in the liver.[ In LC patients, hepatocytes require the consumption of more adenosine-5′-triphosphate (ATP) to overcome the disease. When there is a shortage in ATP supply, liver function deteriorates. In patients with LC status, oxygen will be consumed more quickly to overcome disease, which will result in a shortage of oxygen for ATP synthesis in hepatocytes. Thus, hepatocytes activate the glycolysis pathway, which does not require oxygen to produce ATP.[ These may well explain why LC patients with higher npRQ value have favorable clinical outcomes.

Higher eGFR value was a favorable independent predictor for all cases, as indicated by indirect calorimetry, in our multivariate analyses. Renal function in LC patients is of importance prognostically.[ Indeed, the model for end-stage liver disease includes serum creatinine level.[ Conversely, higher BMI was associated with better clinical outcome in our analysis. Furthermore, in this study, patients with BMI ≥25 kg/m2, which defines obesity in Japan, had significantly better survival than those without this BMI level across all patients (P = 0.0023) and for patients without HCC, as indicated by indirect calorimetry (P = 0.0106).[ Our data indicate that LC patients with obesity can produce favorable clinical outcomes, although sarcopenic obesity is growing to have a significant role as an adverse predictor due to the double metabolic burden derived from excess adiposity (obesity) and low muscle mass (sarcopenia).[ LC patients can experience complications of sarcopenia.[ Although examining the effect of sarcopenic obesity on survival is beyond the scope of our current analysis, this clinical entity will attract attention in the future.

In our results, the serum albumin level and PT in patients with npRQ ≥0.85 were significantly higher than those in patients with npRQ <0.85. One possible reason for these results is that more advanced LC status can be easily complicated with PEM.[ Conversely, it is also of note that the prevalence of HCC in patients with npRQ ≥0.85 (23.0%, 28/122) was almost identical to that in patients with npRQ <0.85 (21.3%, 26/122). Therefore, the presence of HCC may not affect the npRQ value.

For the entire cohort, 74 patients (30.3%) died during the follow-up period, while in HCV-related LC patients with SVR (n = 50), only 3 patients died. Achievement of SVR could have strong survival impact.[ In these patients, the effect of the npRQ value may diminish.

We acknowledge several limitations to this study. First, evaluation for extrahepatic shunts and blood ammonia levels potentially affecting prognosis was not performed in our analysis.[ Second, this is a retrospective observational study. Third, a considerable number of subjects were excluded from our analysis because of a loss of follow-up within a short period of time after performing indirect calorimetry, thereby resulting in bias. Fourth, measurement of the npRQ value using indirect calorimetry is not easy to perform in daily clinical practice due to its high cost. Thus, our results may be unable to be applied to patients in other institutions that do not have this equipment. However, our current results demonstrated that the npRQ value is associated with clinical outcomes in LC patients.

In conclusion, the npRQ value, as assessed by indirect calorimetry, can be helpful for predicting outcomes for LC patients even in the era of established nutritional interventional treatments for LC. In patients with lower npRQ value, adequate interventions may be required.