Oncological prognosis and morbidity of hepatectomy in elderly patients with hepatocellular carcinoma: a propensity score matching and multicentre study

Purpose With increasing life expectancy, the number of elderly patients (≥ 65 years) with hepatocellular carcinoma (HCC) has steadily increased. Hepatectomy remains the first-line treatment for HCC patients. However, the prognosis of hepatectomy for elderly patients with HCC remains unclear. Methods Clinical and follow-up data from 1331 HCC patients who underwent surgery between 2008 and 2020 were retrospectively retrieved from a multicentre database. Patients were divided into elderly (≥ 65 years) and non-elderly (< 65 years) groups, and PSM was used to balance differences in the baseline characteristics. The postoperative major morbidity and cancer-specific survival (CSS) of the two groups were compared and the independent factors that were associated with the two study endpoints were identified by multivariable regression analysis. Results Of the 1331 HCC patients enrolled in this study, 363 (27.27%) were elderly, while 968 (72.73%) were not. After PSM, 334 matched samples were obtained. In the propensity score matching (PSM) cohort, a higher rate of major morbidity was found in elderly patients (P = 0.040) but the CSS was similar in the two groups (P = 0.087). Multivariate analysis revealed that elderly age was not an independent risk factor associated with high rates of major morbidity (P = 0.117) or poor CSS (P = 0.873). The 1-, 3- and 5-year CSS rates in the elderly and non-elderly groups were 91.0% versus 86.2%, 71.3% versus 68.8% and 55.9% versus 58.0%, respectively. Preoperative alpha fetoprotein (AFP) level, Child‒Pugh grade, intraoperative blood transfusion, extended hemi hepatectomy, and tumour diameter could affect the postoperative major morbidity and preoperative AFP level, cirrhosis, Child‒Pugh grade, macrovascular invasion, microvascular invasion (MVI), satellite nodules, and tumor diameter were independently and significantly associated with CSS. Conclusion Age itself had no significant effect on the prognosis of elderly patients with HCC after hepatectomy. Hepatectomy can be safely performed in elderly patients after cautious perioperative management. Supplementary Information The online version contains supplementary material available at 10.1186/s12893-023-02230-0.

Oncological prognosis and morbidity of hepatectomy in elderly patients with hepatocellular carcinoma: a propensity score matching and multicentre study Introduction Hepatocellular carcinoma (HCC) ranks third among causes of cancer-related deaths worldwide [1] and ranks second in men [2].HCC has a poor prognosis, with a 1-year survival rate of less than 50% after disease onset [3].The incidence of HCC has increased dramatically in the last three decades and the number of HCC cases is expected to continue increasing in the next 20 years [4].With the rapid development of medical technology and increased access to health care, as well as improved human life expectancy, the ageing of the population is an inevitable phenomenon [5].Ageing is inherently a common risk factor for the development of malignant tumours, and HCC has a higher specific incidence among elderly patients worldwide [6].Studies have shown that the development of HCC is closely associated with ageing [7], and the number of elderly patients with HCC is expected to increase considerably [8].
Currently, partial hepatectomy remains the preferred treatment for selected patients with HCC [9].Previous studies have revealed that hepatectomy can also be safely performed in elderly HCC patients [10,11].However, the potential benefits among this population are not well elucidated.In addition, identification of risk factors associated with tumour prognosis in elderly patients, more accurate assessment of surgical risk, and more comprehensive perioperative management are essential for reducing the probability of serious postoperative complications and improving long-term prognosis [12].Studies have shown that age is strongly associated with the longterm prognosis of some tumours.For example, in gastric [13], colorectal [14] and lung [15] cancers, the long-term prognosis is worse in younger patients, while in prostate [16] and thyroid [17] cancers, the prognosis is worse in older patients.Age can act as an independent factor affecting long-term prognosis.There are few studies on the correlation between age and long-term prognosis in HCC patients treated with hepatectomy, and the findings are varied and contradictory [18].Some studies have concluded that elderly patients have similar long-term outcomes as younger patients, and sometimes elderly patients could even perform better in recurrence [19,20], however these studies have not revealed the relationship between old age and prognosis.Another study concluded that young (≤ 35 years of age) HCC patients had a higher recurrence rate than elderly patients, but overall survival was similar in both young and old patients.This may be due to the higher incidence of chronic disease and nononcologic death in elderly patients, but the study neglected the relatively large middle-aged patient population (> 35 years and < 65 years of age) [21].Such studies also have the disadvantage of being mostly singlecentre studies with small sample sizes, wherein some selection bias is inevitable.
This study aimed to use a multicentre database to compare cancer-specific survival (CSS), and the probability of serious postoperative complications in elderly versus nonelderly HCC patients treated with R0 hepatectomy, and to use the PSM method to balance baseline characteristics between the two datasets.We defined cancerspecific survival as the time from diagnosis of HCC to death due to HCC.

Study population
The data used in this study were collected between 2008 and 2020 from patients who underwent curative-intent hepatectomy for HCC at the First, and Second Affiliated Hospital of Anhui Medical University and, the First Affiliated Hospital of University of Science and Technology of China, which are three high-volume hepato-pancreato-biliary surgery centres.The study was conducted in accordance with the Declaration of Helsinki and Ethical Guidelines for Clinical Studies (No. P-2021-1230).This was a retrospective clinical study, and the family members of the participants were informed of the scientific use of their clinical data at the time of the preoperative interview and were not required to provide a separate written informed consent.Radical hepatectomy was defined as the absence of residual tumour on intraoperative visual or ultrasound examination and the absence of residual tumour on postoperative microscopic pathological examination of the resection margins.According to the World Health Organization, most developed countries have adopted the definition of ≥ 65 years as elderly individuals, that is, elderly patients were defined as ≥ 65 years of age at the time of the procedure and nonelderly patients were defined as < 65 years of age at the time of the procedure.The exclusion criteria for this study were R1 or R2 resection (palliative surgery), recurrent HCC, preoperative adjuvant therapy, combined hepatocellularcholangiocarcinoma or other malignancies, and a significant lack of clinical or follow-up data.

Data collection
All operations were performed by experienced liver surgeons.The required clinical data were divided into three parts, namely preoperative examination, surgical variables, and tumour characteristics.Preoperative examinations included age, sex, American Society of Anaesthesiologists (ASA) score, body mass index (BMI), comorbidity, alpha-fetoprotein (AFP), alanine aminotransferase (ALT), aspartate transaminase (AST), hepatitis B virus (HBV) infection, cirrhosis, portal hypertension, Barcelona Clinic Liver Cancer (BCLC) staging and Child-Pugh grade.Surgical variables included operative approach, margin width, blood loss, intraoperative blood transfusion, operation duration, anatomical hepatectomy, extended hemi hepatectomy and occlusion.Tumour characteristics included macrovascular invasion, microvascular invasion (MVI), satellite nodules, tumor diameter, tumour differentiation and tumour envelope.Cirrhosis was confirmed by preoperative imaging, intraoperative exploration, or postoperative pathology.Anatomical hepatectomy was defined as hepatectomy with the systematic removal of a hepatic segment, and segmentation was based on liver function [22].The degree of extended hemi-hepatectomy was defined as more than a hemihepatic lobe that was resected [23].The ASA scores of all patients were determined by a dedicated anaesthesiologist before surgery based on the preoperative findings, and all the ratings were accurate and reliable.

Follow-up
Follow-up was performed primarily through outpatient clinic visits or by telephone.The follow-up included monthly reexamination for 3 months after surgery, once every 3 months at 4-24 months after surgery, every 6 months at 2-5 years after surgery, and annual visits starting at 5 years after surgery.Follow-up evaluation included assessment for postoperative complications, AFP, liver function tests, and imaging tests.Ultrasound, computed tomography (CT), or MRI were considered acceptable but should have included an enhanced CT or MRI within one year.Three follow-up endpoints were defined: presence of serious complications, tumour recurrence, and patient death.Major morbidity was defined as postoperative complications with Clavien-Dindo grade ≥ 3 related to the procedure, including perioperative complications or readmission due to complications.Tumour recurrence must be confirmed by enhanced CT, MRI, positron emission tomography (PET)-CT, or needle biopsy, with detailed documentation of the time of recurrence, AFP value, and site of recurrence.It is essential to confirm whether patient death is cancer specific.Noncancer specific death includes death due to liver failure graderelated complications, gastrointestinal bleeding, unexplained death due to age, and other tumour induced death.

Statistical analysis
Patients in the elderly group and non-elderly group were matched by using the PSM method as described by Rosenbaum and Rubin [24,25].The propensity score for an individual was estimated using logistic regression [26] and the co-variables in the regression model included sex, ASA score, BMI, comorbidity, AFP, ALT, AST, HBV infection, cirrhosis, portal hypertension, BCLC staging, Child-Pugh grade, operative approach, margin width, blood loss, intraoperative blood transfusion, operation duration, anatomical hepatectomy, extended hemi hepatectomy, occlusion, macrovascular invasion, MVI, satellite nodules, tumor diameter, tumour differentiation and tumour envelope.We applied 1:1 nearest neighbor matching without replacement in order to ensure that conditional bias was minimized.The nearest neighbor matching was based on a greedy matching algorithm [27], which matched each unit in the elderly group to a unit in the non-elderly group that had the closest propensity score.The appropriateness of matching was assessed by comparing the standardized differences in covariate means for continuous variables and differences in covariate means for dichotomous variables for the matched and unmatched samples.Small absolute values in standardized differences (< 10%) were assumed to support the assumption of balance between the treatment groups.A calliper width of 0.020 resulted in the best trade-off between homogeneity and retained sample size.More stringent calliper was also attempted but 0.020 gave the best matching model.
Categorical variables are expressed as numbers or proportions, and continuous variables are expressed as the mean ± standard deviation or median (range).The t test was used to compare continuous variables, and the chisquare test was used to compare categorical variables.Survival analysis was performed using the Kaplan-Meier method, and the log-rank test was used to analyse differences between the two groups.Univariate and multivariate analysis was used to identify prognostic variables associated with CSS and major morbidity.All statistical analyses were conducted using SPSS 23.0 (SPSS Inc, Armonk, NY, USA) and R 4.2.0 (http://www.r-project.org/, R Development Core Team).A P value < 0.05 was considered to indicate a significant difference in a 2-tailed test.

Clinicopathological characteristics
Using the inclusion and exclusion criteria, a total of 1331 patients were enrolled in the present study, including 363 (27.27%) elderly patients with a mean age of 70.38 ± 4.46 years (P < 0.001), and 968 (72.73%) nonelderly patients with a mean age of 51.55 ± 8.60 years (P < 0.001) (Fig. 1).Data on the 334 matched groups of elderly and nonelderly patients were obtained by PSM.In the matched groups, the mean age of elderly patients was 70.25 ± 4.47 years (P < 0.001), and that of nonelderly patients was 52.16 ± 8.66 years (P < 0.001).The clinical and pathological characteristics and surgical variables between the two groups before and after PSM are showed in Table 1.After PSM, there were no significant differences in any liver or tumour-related variables between the nonelderly and elderly patients (all P > 0.200), except for the ASA score (34.13% vs. 20.36%,P < 0.001) and comorbidities (48.50% vs. 14.67%,P < 0.001).This is consistent with our common sense of life, as the ASA score and comorbidities are intrinsic variables that are related to age.The love plots of the absolute standardized differences before and after PSM show that all covariates are below the threshold of 10% except the ASA score and comorbidities.This means that propensity score matching has balanced the treatment and control groups regarding these covariates (Fig. 2).

Comparisons of short-and long-term oncological prognosis
As shown in Table 2, the risk of major morbidity in the elderly group was higher than that in the nonelderly group, both before (14.60% vs. 7.64%, P < 0.001) and after (14.07%vs. 6.89%,P = 0.004) PSM.The rate of perioperative death in the elderly group was higher than that in the nonelderly group before PSM (4.96% vs. 1.76%,P = 0.020).However, this rate was not significantly different between the two groups after PSM (2.40% vs. 1.80%,P = 0.787).These results indicated that the elderly group suffered a higher risk of serious postoperative complications, but old age did not increase the probability of perioperative death.

Discussion
The present study utilized a multicentre database to describe and compare the short-and long-term outcomes of elderly and nonelderly HCC patients who underwent R0 resection.The sample size was large with comprehensive data, and the three hospitals selected for the study were large, high-volume hepato-pancreatobiliary surgery centres with accurate and reliable clinical data.PSM was performed before the analysis and comparison, to balance the differences in baseline characteristics between the two datasets and to reduce selection bias [27].After PSM, the data from the two groups did not differ significantly and were comparable.Multivariate analysis was used to compare the effects of age on the short-and long-term outcomes of HCC patients after surgery.The matched sample size (n = 334) was large, which improved the accuracy and credibility of the results.
The data from both groups showed that the elderly group had a higher probability of serious postoperative complications but the rates of death during the perioperative period and follow-up were similar after PSM.Elderly patients tended to have more comorbid chronic diseases (48.76% vs. 18.29%,P < 0.001) and are in worse physical condition.Cirrhosis and its associated portal hypertension (38.64% vs. 33.13%,P = 0.014) are more severe in elderly patients than in nonelderly patients, and progression of liver fibrosis results in worse prognosis [28].In addition, factors associated with cancer in elderly patients such as sarcopenia [29], cachexia [30], and malnutrition [31] also reduce the safety of surgery and survival of elderly patients to some extent.Elderly patients have higher noncancer-specific mortality than nonelderly patients, which adversely affects survival.On the other hand, other studies have shown that younger patients have more malignant and aggressive tumours, a Fig. 3 Kaplan-Meier curves comparing the CSS of elderly and nonelderly patients with HCC in the propensity score-matched cohort (P = 0.087) higher probability of recurrence, and a greater tendency towards metastasis [32], consistent with the findings of the present study.Meanwhile, the preoperative evaluation of elderly patients was more rigorous, the surgical treatment of nonelderly patients was more aggressive, and more nonelderly patients with advanced disease (BCLC classification B or C) underwent radical hepatectomy (37.60% vs. 30.03%,P = 0.012).This may be related to the llower recurrence and cancer-specific death in the elderly group during follow-up.Meanwhile, we found an interesting result in multivariate analysis that high AFP have a protective role towards morbidity (OR = 0.48), and this may also be related to the above differences.In our study, more nonelderly patients with higher malignancy and later stage underwent hepatectomy and better physical condition meant that this group had a lower probability of major morbidity after hepatectomy than elderly patients.It is well known that AFP is significant risk factor for tumor aggressiveness [33,34], which may be the reason why high AFP levels seem to have a protective effect on major complications in the present study.
Based on the current patterns of population ageing, the progressive increase in life expectancy has resulted in an increased age of onset of HCC and the number of elderly HCC patients is expected to increase substantially in the future [8].Thus, prognostic analysis of elderly HCC patients is expected to become a future focus.However, poorer perioperative prognosis and higher noncancer-specific death also indicate that for elderly postoperative HCC patients, not only should the detection and treatment of tumour recurrence be emphasized, but the overall physical condition of the patient warrants attention.Also, the prognosis of the patient can be improved through comprehensive monitoring and treatment.Regarding the age of the patients, especially younger patients (< 35 years old), postoperative monitoring of HCC recurrence must be improved because of the higher probability of recurrence [21].Previous studies have used overall survival (OS) to evaluate long-term oncologic surgical outcomes.However, OS focuses only on overall patient survival and does not consider whether the tumour was the cause of reduced life expectancy, whereas CSS has been used to analyse tumourspecific mortality [35].In the present study, we considered CSS a more representative and accurate reference indicator for assessing long-term outcomes of surgical treatment with more clinical guiding significance.Elderly patients were found to have similar CSS as nonelderly patients, and radical partial hepatectomy is still considered the preferred and most beneficial treatment option for earlystage elderly HCC patients [36].
The present study has some limitations.First, it was a retrospective clinical study with inherent selection bias, and the use of PSM reduced but did not completely eliminate this bias.Second, the patient population in this study was entirely Chinese with a high proportion of HBV infected patients, which differs from alcoholic liver disease, which is more prevalent in Western countries.These potential confounding factors, which may lead to differences in the study findings, have not yet been verified.Third, the present study confirmed that elderly patients have a higher probability of serious postoperative complications, which affects their overall survival, but the type and severity of the complications were not further investigated [37][38][39].Fourth, due to excessive variability, only chronic diseases were included as a separate overall variable in the present study, and no separate analysis of the type and impact of chronic diseases was performed.Although this is a multicentre study, prospective studies are still needed to obtain more reliable conclusions.

Conclusion
The present study showed that elderly HCC patients undergoing hepatectomy can achieve similar oncological outcomes, to those of nonelderly patients and that partial hepatectomy remains an effective and beneficial treatment option of choice for elderly HCC patients.

Fig. 2
Fig. 2 Absolute standardized difference before and after PSM

Table 1
The clinical, operative and pathological characteristics between the elderly and nonelderly group before and after propensity score matching Values in parentheses are percentages unless stated otherwise; #values are mean(s.d.).AFP, alpha-fetoprotein; ASA, American Society of Anesthesiologists; ALT, alanine aminotransferase; AST, aspartate transaminase; BCLC, Barcelona Clinic Liver Cancer; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; MVI, microvascular invasion Table2Comparisons of short-and long-term oncological outcomes between the elderly and the nonelderly groups before and after propensity score matchingCharacteristicsBefore PSM(n = 1331) P After PSM(n = 668)

Table 3
Univariate and multivariate logistic regression analyses for predicting postoperative major morbidity after partial hepatectomy for hepatocellular carcinoma after PSM.