Introduction
Taiwan is well known for endemic viral hepatitis. The prevalence of chronic hepatitis B virus (HBV) infection is about 5 per cent worldwide, but 15–20 per cent in Taiwan. In contrast, hepatitis C virus (HCV) infection is more prevalent in Western countries. Oral cancer is one of the four most common cancer types in Taiwanese males. 1 Many patients with head and neck cancer may have either underlying hepatitis-induced liver cirrhosis or alcohol abuse related liver disease.Reference Perkins and Perkins 2 This inspired our interest in investigating the correlation between liver cirrhosis and oral cancer.
Liver disease requiring surgery has been proven to increase the risk of morbidity and mortality.Reference Friedman 3 – Reference Friedman 5 The severity and the type of liver disease is a strong determinant of the level of risk for peri-operative complications. Child–Turcotte–Pugh classification and Model for End-stage Liver Disease score are the two most commonly used measures of liver function in cirrhosis, and are known to be prognostic factors influencing surgical risk.Reference Farnsworth, Fagan, Berger and Awad 6 – Reference Hanje and Patel 9 Cardiac surgery,Reference Klemperer, Ko, Krieger, Connolly, Rosengart and Altorki 10 – Reference Shaheen, Kaplan, Hubbard and Myers 12 spinal surgery,Reference Liao, Chen, Chen, Niu, Fu and Lai 13 hepatic surgeryReference Wu, Yeh, Lin, Liu and P'Eng 14 – Reference Clavien, Petrowsky, DeOliveira and Graf 16 and other intra-abdominal proceduresReference Demetriades, Constantinou, Salim, Velmahos, Rhee and Chan 17 increase the risk of post-operative mortality or complications in patients with cirrhosis. Some studies have indicated that alcoholic liver cirrhosis may induce cellular changes and later increase the risk of developing oral cancer.Reference Lekholm and Stenman 18 , Reference Randi, Altieri, Gallus, Franceschi, Negri and Talamini 19 One retrospective review mentioned the association between post-operative complications, mortality and liver cirrhosis following surgical ablation plus microsurgical free tissue transfer.Reference Kao, Chang, Ching, Tsao, Cheng and Wei 20 Another retrospective study identified liver cirrhosis as the most common factor contributing to non-treatment related mortality in patients receiving oral cancer treatment.Reference Chang, Yeh and Huang 21 So far, no cohort study has been conducted on the mortality rates of oral cancer patients with different liver conditions.
The central government of Taiwan introduced the National Health Insurance system in 1995 to ensure its citizens equal access to healthcare. Since then, a nationwide cohort database has been established. Our aim was to assess the relationship between the severity of liver cirrhosis and risk of death from oral cancer after surgery using a cohort from the Taiwan National Health Insurance Research Database.
Materials and methods
Ethical considerations
The protocol of this study was approved by the Institutional Review Board of the Dalin Tzu Chi General Hospital, Taiwan. Written informed consent was not necessary, as identification numbers and personal information had been removed from the dataset prior to data analysis.
Patients and study design
We used data from the National Health Insurance Research Database, which covers more than 99 per cent of the Taiwanese population. Patients who were diagnosed as having oral cancer (surgically removed between 2008 and 2011) were included. A total of 3583 patients were identified. Deaths were confirmed by searching the National Register of Deaths Database of Taiwan. The patient characteristics recorded included age, gender, underlying diseases (e.g. hypertension, diabetes, hyperlipidaemia), socioeconomic status (low, moderate or high) and geographical region (northern, central, southern or eastern). The evaluation of socioeconomic status was based on income in Taiwan and several urbanisation variables.Reference Chang, Chang, Hsu, Lin, Lai and Su 22
The cohort was identified by querying the Taiwan National Health Insurance Research Database using the International Classification of Diseases, Ninth Revision (‘ICD-9’) codes 140–149 for oral cancer; codes 142 (malignant neoplasm of the major salivary glands) and 147 (malignant neoplasm of the nasopharynx) were excluded.
The patients were identified from the record as having cirrhosis using the International Classification of Diseases, Ninth Revision codes ranging from 571.2 (alcoholic cirrhosis of the liver) to 571.5 (cirrhosis of the liver without mention of alcohol). The patients had been diagnosed as having cirrhosis within the three years prior to the diagnosis of oral cancer, and visited an out-patient facility at least three times or were admitted to a ward at least once. Decompensated liver cirrhosis included codes: 572.3 (portal hypertension); 578 (haemorrhage of the gastrointestinal tract); 789.5 (ascites); 572.2 (hepatic coma); 070.0, 070.2, 070.4 and 070.6 (viral hepatitis with hepatic coma); 572.4 (hepatorenal syndrome); 511 (pleural effusion); and 572.8 (other sequelae of chronic liver disease).
Statistical analysis
SPSS software (version 15; SPSS, Chicago, Illinois, USA) was used for data analysis, and a p-value of < 0.05 was used to determine statistical significance. The impact of cirrhosis on oral cancer survival rates was estimated with Kaplan–Meier survival curves, and differences were assessed by means of the log-rank statistic. The chi-square test was used to compare categorical independent variables. A multivariable Cox hazards regression model was used to estimate hazard ratios, with 95 per cent confidence intervals (CIs), of death and different risk factors.
Results
A total of 3583 patients with oral cancer diagnosed from 2008 to 2011 and treated by head and neck surgery were identified in the National Health Insurance Research Database. Their basic characteristics, including age, gender, underlying diseases, socioeconomic status and geographical region, and the numbers who received chemotherapy or radiotherapy, are shown in Table I.
Baseline characteristics*
Data represent numbers of patients (and percentages), except where indicated otherwise. *Total n = 3583. SD = standard deviation
Of the 3583 patients, 3471 (96.9 per cent) had no liver cirrhosis, while 72 patients (2.0 per cent) had liver cirrhosis without decompensation and 40 patients (1.1 per cent) had liver cirrhosis with decompensation. The mean age (± standard deviation) of patients in these three groups was 52 ± 10, 53 ± 10 and 51 ± 9 years, respectively. Differences between the three groups in terms of frequency of underlying diseases, including hypertension, myocardial infarction, congestive heart failure, liver disease, cardiovascular disease, chronic obstructive pulmonary disease, chronic kidney disease, diabetes and peripheral vascular disease, were analysed. Only the frequency of liver disease differed significantly between the groups (3.9 per cent, 79.2 per cent and 92.5 per cent, respectively; p < 0.001).
Oral cancer patients with decompensated liver cirrhosis incurred the highest mortality rate (Figure 1). The number of deaths and mortality rates for the three groups are shown in Table II. The mortality rates in the normal liver function, cirrhosis without decompensation and cirrhosis with decompensation groups were similar at one month (0.2 per cent, 0.0 per cent and 0.0 per cent; p = 0.908), but significantly different at six months (4.8 per cent, 9.7 per cent and 15.0 per cent; p = 0.002) and one year (14.8 per cent, 20.8 per cent and 37.5 per cent; p < 0.001).
Cumulative risk for oral cancer patients stratified by liver cirrhosis status (p < 0.001).
Mortality among oral cancer patients from 2008 to 2011*
Data represent numbers of patients (and percentages). Mortality rates in the normal liver function, cirrhosis without decompensation and cirrhosis with decompensation groups were similar at one month (p = 0.908), but significantly different at six months (p = 0.002) and one year (p < 0.001). *Total n = 3583; † n = 3471; ‡ n = 444; **n = 397
The adjusted hazard ratios for mortality in the surgically treated oral cancer patients are listed in Table III. The adjusted hazard ratios for post-operative mortality in the non-decompensated and decompensated cirrhosis groups versus the normal liver function group were, respectively, 2.61 (95 per cent CI = 1.05–6.51; p = 0.039) and 5.29 (95 per cent CI = 1.84–15.21; p = 0.002) at six months, and 2.01 (95 per cent CI = 1.11–3.64; p = 0.021) and 4.84 (95 per cent CI = 2.55–9.17; p < 0.001) at one year.
Multivariate adjusted hazard ratios of mortality among oral cancer patients from 2008 to 2011
*Adjusted for patients' age, gender, chemotherapy, radiotherapy, hypertension, myocardial infarction, congestive heart failure, liver disease, cardiovascular disease, chronic obstructive pulmonary disease, chronic kidney disease, diabetes, peripheral vascular disease, socioeconomic status and geographical regions. HR = hazard ratio; CI = confidence interval
Those who received chemotherapy had a hazard ratio of 1.99 (95 per cent CI = 1.46–2.71; p < 0.001) at six months and 2.65 (95 per cent CI = 2.22–3.17; p < 0.001) at one year. However, radiotherapy was not a significant risk factor, with a hazard ratio of 0.90 (95 per cent CI = 0.51–1.60; p = 0.741) at six months and 0.91 (95 per cent CI = 0.66–1.24; p = 0.561) at one year.
The analysis revealed other significant factors. The hazard ratio for hypertension was 0.75 (95 per cent CI = 0.60–0.93; p = 0.012) and for congestive heart failure was 1.66 (95 per cent CI = 1.23–2.24; p = 0.001) at one year. In comparison to low socioeconomic status, high socioeconomic status had a reduced hazard ratio of 0.58 (95 per cent CI = 0.37–0.92; p = 0.021) at six months and was 0.74 (95 per cent CI = 0.58–0.93; p = 0.013) at one year. Patients residing in southern or eastern geographical regions of Taiwan had a hazard ratio of 0.78 (95 per cent CI = 0.65–0.94; p = 0.009) compared to those residing in the northern or central regions.
Discussion
We found that liver dysfunction at the time of oral cancer diagnosis can predict mortality. The level of liver function (normal, non-decompensated cirrhosis or decompensated cirrhosis) can be used to predict six-month and one-year mortality rates in those who received surgical treatment. Significant differences in mortality between the normal liver function, cirrhosis without decompensation and cirrhosis with decompensation groups were observed at six months (4.8 per cent, 9.7 per cent and 15.0 per cent) and one year (14.8 per cent, 20.8 per cent and 37.5 per cent). In the present study, chemotherapy (neoadjuvant or adjuvant), but not adjuvant radiotherapy, was found to significantly increase mortality risk.
Some of the strengths of our study include the nationwide population-based cohort design, the large sample population (n = 3583) and the availability of complete follow-up information from the Taiwan National Health Insurance Research Database. To our knowledge, the only evidence of a positive association between hepatitis C infection and oral cavity cancer was provided by a cohort study from Taiwan.Reference Su, Chang, Chen, Sung, Huang and Chiou 23 Another cohort study from the UK showed that oral cancer risk was elevated in alcoholic cirrhosis and other liver diseases.Reference Goldacre, Wotton, Yeates, Seagroatt and Collier 24 Nevertheless, no cohort study has assessed mortality risk in oral cancer patients with different levels of liver dysfunction. Furthermore, our study adopted an alternative viewpoint. The Child–Turcotte–Pugh classification and Model for End-stage Liver Disease score are generally used as prognostic indicators of surgical risk in cirrhotic patients.Reference Farnsworth, Fagan, Berger and Awad 6 – Reference Hanje and Patel 9 In addition, cardiac surgery,Reference Klemperer, Ko, Krieger, Connolly, Rosengart and Altorki 10 – Reference Shaheen, Kaplan, Hubbard and Myers 12 spinal surgery,Reference Liao, Chen, Chen, Niu, Fu and Lai 13 hepatic surgeryReference Wu, Yeh, Lin, Liu and P'Eng 14 – Reference Clavien, Petrowsky, DeOliveira and Graf 16 and other intra-abdominal surgeryReference Demetriades, Constantinou, Salim, Velmahos, Rhee and Chan 17 increase the risk for post-operative mortality or complications in cirrhotic patients. In our study, rather than using Child–Turcotte–Pugh or Model for End-stage Liver Disease score, we used liver decompensation, which is easily and rapidly established from a survey of database records, to classify patients.
Despite vaccinating all neonates against HBV since 1984, which has markedly reduced hepatitis B surface antigen (HBsAg) seroprevalence in the younger generation, Taiwan remains an HBV-hyperendemic region, with more than three million carriers. Chronic infections with HBV and HCV are well-established causes of cirrhosis. Globally, 57 per cent of cirrhosis is attributable to either HBV (30 per cent) or HCV (27 per cent),Reference Perz, Armstrong, Farrington, Hutin and Bell 25 but this percentage may be far higher in Taiwan. Nearly 80–90 per cent of liver cirrhosis cases are a direct consequence of chronic HBV infection according to an early Taiwanese survey.Reference Sung, Chen and Lee 26 In addition, chronic alcohol consumption not only causes oral cavity cancer, but it also accelerates cirrhotic change and hepatocarcinogenesis in HBV infection.Reference Poschl and Seitz 27
Cirrhosis refers to the replacement of healthy nodules with scarred or regenerative nodules of liver parenchyma, leading to impaired liver function. If the underlying cause is eliminated, pathophysiological changes are still reversible. When non-decompensated cirrhosis progresses to decompensated cirrhosis, liver functions deteriorate more rapidly and patients will suffer from more complications. However, the causes of higher mortality risk in these patients are not just based on hepatic failure. An Italian case–control study found that liver cirrhosis itself can raise the risk of developing several types of malignancy, such as oral cancer, pharyngeal cancer, oesophageal cancer, primary liver cancer and other gastrointestinal tract cancers.Reference Randi, Altieri, Gallus, Franceschi, Negri and Talamini 19 A Danish cohort study examining the risks of cirrhosis-related and non-cirrhosis-related death found an apparent synergy between cirrhosis and co-morbidity effects on mortality.Reference Jepsen, Vilstrup, Andersen, Lash and Sorensen 28 All these factors may contribute to higher mortality rates.
There are several ways to reduce the mortality risk of cirrhotic patients. First, physicians making the initial malignancy diagnosis need to take a complete substance abuse history and perform a complete assessment of liver function. Patients who are alcoholic should be encouraged to abstain from alcohol, and those who are viral hepatitis carriers should be treated with antiviral agents. Both chemotherapy and steroids may induce HBV reactivation. Nucleoside analogues such as entecavir, telbivudine, lamivudine and adefovir have been extensively used for the treatment of HBV carriers. They can lower viral load in blood, and inhibit viral replication and viral spread to non-infected hepatocytes. However, resistance to these agents, especially lamivudine and adefovir, is a major concern. Entecavir was shown to be more potent than lamivudine in both hepatitis B envelope antigen (HBeAg)-positive and HBeAg-negative patients.Reference Chang, Gish, Man, Gadano, Sollano and Chao 29 , Reference Lai, Shouval, Lok, Chang, Cheinquer and Goodman 30 Tenofovir is another nucleoside analogue, initially indicated for treating human immunodeficiency virus, which was later found to have an antiviral effect on HBV. Tenofovir and entecavir (which are associated with lower complication rates, greater suppression of HBV DNA levels and smaller increases in Child–Turcotte–Pugh scores) are currently the most effective and safest nucleoside analogues for long-term use in HBV carriers.Reference Köklü, Tuna, Gülşen, Demir, Aydin and Muhammet 31
Second, some liver-protecting agents may help delay deterioration of liver functions. Belonging to the order Aphyllophorales and the family Polyporaceae, Antrodia cinnamomea (or Antrodia camphorata, Taiwanofungus camphoratus, commonly called ‘niu-chang-chih’ in Chinese) is a valuable, perennial mushroom native to Taiwan. It is commonly used as a liver protective, anticancer, antioxidant, anti-inflammatory, antihypertensive and immunomodulatory agent.Reference Ao, Xu, Lu, Xu, Zhang and Dou 32 , Reference Hseu, Chang, Hseu, Lee, Yech and Chen 33 The mechanisms underlying its activities are mainly free radical scavenging,Reference Lin, Kuo, Lin, Fang and Wang 34 retardation of liver fibrosis progressionReference Shu and Lung 35 and induction of hepatocyte regeneration. Extracts of A cinnamomea are easily obtainable in Taiwan. Silibinin (also known as silybin) is another hepatoprotector that shows an ability to significantly reduce liver-related toxicity and mortality.Reference Haddad, Vallerand, Brault and Haddad 36 , Reference Al-Anati, Essid, Reinehr and Petzinger 37 It is a major constituent of silymarin, and is extracted from a complex of flavonoids from milk thistle seeds. The mechanisms are complex, but it mainly blocks the penetration of toxins into hepatocytes and prevents apoptosis due to intracellular free radical mediated DNA damage.Reference Kostek, Szponar, Tchórz, Majewska and Lewandowska-Stanek 38 Many other hepatoprotectors are more or less effective.
Third, surgeons and oncologists should carefully assess patients' physiological condition and medical co-morbidities, so that surgery, chemotherapy or radiotherapy can be provided efficaciously and safely. The effect of interactions between cirrhosis and co-morbidities on mortality should not be overlooked. It has been demonstrated that oral cancer patients with liver cirrhosis have increased risk of developing secondary malignancy, such as pharyngeal cancer, oesophageal cancer, primary liver cancer or other visceral cancers.Reference Perz, Armstrong, Farrington, Hutin and Bell 25 Even if oral cancer is managed successfully with a series of curative treatments and periodic follow up, cirrhotic patients with oral cancer should receive appropriate screening measures for other cancers.
-
• Liver cirrhosis status has a significant effect on survival following surgical treatment of head and neck cancer
-
• Chemotherapy enhances mortality risk, while radiotherapy does not
-
• Cirrhotic patients should have a thorough pre-treatment survey and underlying conditions should be managed carefully
Our study has several limitations. First, no clinical studies have compared the survival of liver cirrhotic patients with and without head and neck cancer. We do not know exactly how head and neck cancer affects survival in cirrhotic patients because of the lack of such data. This is a limitation in this cohort. Second, cases of oral cancer, liver cirrhosis and decompensated conditions were identified solely on the basis of International Classification of Diseases, Ninth Revision codes. If there were some innocent omissions in coding, the numbers of cirrhotic and decompensated cirrhotic patients would be underestimated, even if charts were reviewed and patients were interviewed to verify the accuracy. Third, the oral cancer stage, which determines the type and length of treatment, was not recorded in the database. In this study, the mortality rates were estimated for patients with stage I to stage IVc disease as a whole, but not for subgroups based on stage. Fourth, the effect of chemotherapy on the relative risk of different liver dysfunctions was not shown in our study, so whether chemotherapy should be waived in those with worse liver function remains unknown. As risk of death was higher when chemotherapy was used (and lower when radiotherapy was used), it would be interesting to investigate whether treatment with concurrent chemoradiation is hazardous or safe for cirrhotic patients. Fifth, the actual cause of death in oral cancer patients merits further investigation and discussion. This issue is complicated, as multiple factors, including oral cancer progression, synergy of cirrhosis-related complications and co-morbidities, hepatic failure, and secondary cancer, may all be lethal.
Conclusion
We conclude that, in Taiwan, liver function at the time of oral cancer diagnosis can predict prognosis, with a higher risk of death in patients with worse liver function. Chemotherapy should be given cautiously. We recommend aggressive treatment of cirrhosis to reduce mortality risk.
Acknowledgements
We gratefully acknowledge the contributions of all those involved in data collection, consultation and supervision in this study.
