Is Lymphadenectomy Reasonable for Elderly Intrahepatic Cholangiocarcinoma Patients?

Background In this study, we aimed to determine the impact of lymphadenectomy (LND) on clinical outcomes in ICC patients aged ≥ 70 years. Methods Four hundred and three eligible patients diagnosed with ICC who underwent hepatectomy between 2004 and 2019 were enrolled in the Surveillance, Epidemiology, and End Results database. The impact of LND on perioperative mortality and overall survival (OS) as well as the optimal total number of lymph nodes examined (TNLE) was estimated. Results One hundred thirty-nine pairs of patients were matched by propensity score matching. Perioperative mortality was comparable between the LND and non-LND (nLND) groups (0.7% vs. 2.9%, P = 0.367). The median OS in the LND group was significantly longer (44 vs. 32 months, P = 0.045) and LND was identified as an independent protective factor for OS by multivariate analysis (HR 0.65, 95% CI 0.46–0.92, P = 0.014). Patients with the following characteristics were potential beneficiaries of LND: white, female, no/moderate fibrosis, tumor size > 5 cm, solitary tumor, and localized invasion (all P < 0.05). TNLE ≥ 6 had the greatest discriminatory power for identifying lymph node metastasis (area under the curve, 0.704, Youden index, 0.365, P = 0.002). Patients with pathologically confirmed lymph node metastasis are likely to benefit from adjuvant therapy (40 months vs. 4 months, P = 0.052). Conclusions Advanced age (≥ 70 years) was not a contraindication for LND, which facilitates accurate nodal staging and guides postoperative management. Appropriately selected elderly populations could benefit from LND. Supplementary Information The online version contains supplementary material available at 10.1007/s11605-023-05846-y.


Background
Primary liver cancer ranks as the sixth most commonly diagnosed cancer and the third leading cause of cancer-related death worldwide. 1Intrahepatic cholangiocarcinoma (ICC) is the second most common primary liver cancer with a globally increasing age of incidence. 2The average age at first diagnosis for ICC patients is 70 in the USA. 3 At present, surgical resection remains the mainstay of curative procedures for ICC patients, including elderly patients. 4Due to the high prevalence and extremely negative prognostic effect of lymph node metastasis (LNM) in ICC, 5 routine lymphadenectomy (LND) is recommended by the guidelines of both the European Association for the Study of the Liver (EASL) 6 and the National Comprehensive Cancer Network (NCCN). 7However, the practice rate of LND varies among different centers, 8,9 ranging from 27 to 100%. 10,11 ccording to an international multicenter study (n = 1084), from Qianyi Lin and Jianjun Chen contributed equally to this work and share the first authorship. 1 3   2000 to 2015, the average practice rate of LND was merely 49.4%. 124][15] The low practice rate of LND and conservative LND strategies may be attributed to the increased incidence of complications and the controversial effect on long-term survival. 16,17  was previously indicated that not all ICC patients could benefit from LND. 16,18,19 Furthermore, the potential benefits of LND on long-term survival should be weighed against the risk of postoperative complications related to the procedure itself; elderly ICC patients who underwent complex surgical procedures were more likely to suffer severe complications due to poor performance status, 20,21 and LND may lead to an increase in operative time 22 and blood loss 23 and result in subsequent postoperative morbidity such as bile leakage, hemorrhage, and even perioperative death. 24,25 ether LND is reasonable for elderly (≥ 70 years) ICC patients and the optimal total number of LNs examined (TNLE) remains unclear.Data regarding the impact of LND on clinical outcomes for elderly ICC patients are sporadic.To our knowledge, propensity score matching (PSM) analysis, 26 which could control the selection bias of retrospective studies, has not yet been conducted in such studies.The current study, as such, was designed to compare the perioperative mortality and overall survival of ICC patients aged ≥ 70 years who did or did not undergo LND by PSM and to determine the optimal threshold of TNLE for discriminating LNM using a national database of ICC patients.

Patients and Data Collection
Patients who underwent cancer-directed surgery (codes 20-75, excluding liver transplantation) for ICC between 2000 and 2019 were identified in the SEER database using the 3rd edition International Classification of Disease for Oncology-3 codes by software SEER.Stat 8.4.0.1.Patients with ICC were identified using the primary site code for liver (22.0) and intrahepatic bile duct (22.1) and histology code for malignant neoplasm (8000), malignant tumor cells (8001), carcinoma (8010), undifferentiated carcinoma (8020), adenocarcinoma (8140), and cholangiocarcinoma (8160). 14Patients with recurrent ICC, concomitant or history of other malignancies, distant metastases, performance of preoperative or intraoperative anticancer therapies, age less than 70 years, missing information on LND, or who were lost to follow-up within 30 days after surgery were excluded.In turn, a total of 403 eligible patients were finally included in the analytic cohort.Clinically relevant information including age, sex, race, fibrosis, tumor size, tumor number, histologic grade, vascular invasion, invasive extent, nodal status (categorized as positive, negative, or unstaged), TNLE, and adjuvant therapy was extracted.The extent of LND was not clearly documented in the SEER database, and LND is generally defined as the presence of lymph node examination, as previously reported. 8,12,27,28 Tofurther clarify the clinical significance of TNLE, patients with LND were further classified into adequate-LND (AD-LND) and inadequate-LND (inAD-LND) according to the optimal threshold of TNLE for identifying LNM.The study was performed in accordance with principles of the Declaration of Helsinki and approved by the Institutional Review Board of Zhanjiang Central Hospital, Guangdong Medical University.

Statistical Analysis
Continuous variables are summarized as medians with interquartile ranges (IQRs) and were compared by the Mann-Whitney U test.Qualitative variables were grouped and analyzed by the Pearson chi-squared test or Fisher's exact test, as appropriate.The propensity scores for all patients were estimated by a logistic regression model using preoperative and intraoperative characteristics as covariates: age, sex, race, tumor size, tumor number, and invasive extent.A one-to-one nearest-neighbor matching algorithm with a caliper of 0.1 and without replacement was used.Corresponding standardized mean differences (SMDs) of matched characteristics were estimated.An SMD of < 0.1 indicates very small differences, values between 0.1 and 0.3 indicate small differences, values between 0.3 and 0.5 indicate moderate differences, and values > 0.5 indicate considerable differences. 29Receiver-operating characteristic curve (ROC) analysis was conducted to investigate the optimal threshold of TNLE for detecting LNM.Logistic regression was used to identify factors associated with LNM.Kaplan-Meier curves were used to estimate the median survival of patients who survived the perioperative period (30 days after surgery), and the log-rank test was used to assess differences in overall survival (OS).Potential variables associated with OS were univariately analyzed utilizing Cox proportional hazards regression.Variables with a P value of less than 0.10 in univariate analysis were included in multivariate analysis.Two-tailed P values less than 0.05 were considered statistically significant.Statistical computation was performed by R software (version 4.2.3,R Foundation for Statistical Computing, Vienna, Austria) with packages "MatchIt," "pROC," and "survival."

Baseline Characteristics
The patient enrollment flow diagram is shown in Fig. 1.A total of 403 patients with a median age of 75 years were enrolled in this study, including 183 (45.4%) males and 1 3 220 (54.6%) females.Baseline characteristics are listed in Table 1.Two hundred forty-three patients underwent LND with a median TNLE of three, while 160 patients did not.All preoperative and intraoperative variables were balanced except for the extent of tumor invasion (Table 1).After PSM, 139 pairs of patients were matched, and clinicopathological variables between the two groups were relatively balanced.
After PSM, 155 (56.7%) patients died within the followup period.The median OS was 37 months, with 1-year, 3-year, and 5-year survival rates of 79.1%, 50.9%, and 36.8%,respectively.The median OS in the LND group was significantly longer than that in the nLND group (44 vs. 32 months, P = 0.045; Fig. 3B).After adjusting for sex, tumor number, invasive extent, and nodal status, LND was significantly identified as a protective factor for OS (HR Fig. 2 Receiver operative characteristics (ROC) analysis of identifying the optimal total number of lymph nodes examined (TNLE) for detecting LNM 0.65, 95% CI 0.46-0.92,P = 0.014, Table 3).Interestingly, compared with inAD-LND, no significant survival improvement brought by AD-LND was observed either in the PSM cohort (mOS, 41 vs. 47 months, P = 0.612) or in the populations with negative nodal status (mOS, 40 vs. 55 months, P = 0.402), whereas AD-LND could prolong the OS of patients with LNM (mOS, 56 vs. 9 months, P = 0.043, supplementary 1).

Subgroup Analysis of Overall Survival Stratified by Risk Factors
Subgroup analysis showed that patients with poor status or advanced tumor characteristics, including extremely advanced age (> 80 years), advanced/severe fibrosis, multiple tumors, regional invasion, and vascular invasion, could not benefit from LND (all P > 0.05, Fig. 4).Potential beneficiaries comprised the following characteristics: white, female, no/ moderate fibrosis, tumor size > 5 cm, solitary tumor, and localized invasion (all P < 0.05, Fig. 4).The survival benefit of LND in patients without evidence of vascular invasion was marginally significant (P = 0.074, Fig. 4).

Effects of LND on Postoperative Management
In the matched cohort, patients with unstaged nodal status showed superior OS to those with LNM but inferior OS to those with negative LNs (32 vs. 20 vs. 55 months, P = 0.001, Fig. 5).Receiving adjuvant chemotherapy could potentially improve the OS of patients with LNM (mOS,  6A).Nevertheless, postoperative chemotherapy could not prolong the OS of patients with negative nodal status (94.0 vs. 47.0 months, P = 0.161, Fig. 6B) or unstaged nodal status (33.0 vs. 31.0months, P = 0.801, Fig. 6C).No significant survival benefit was observed with adjuvant radiotherapy regardless of LN status (all P > 0.05, Fig. 6D-F).

Discussion
Whether LND is beneficial for elderly (≥ 70 years) ICC patients who undergo curative-intent hepatectomy, as well as minimal TNLE, remains unclear.The current study controlled baseline characteristics bias by PSM to evaluate the clinical effect of LND on elderly surgery candidates with ICC and demonstrated that an appropriately selected population could benefit from LND without increasing perioperative mortality.TNLE ≥ 6 had the greatest discriminatory power to identify LNM and guide postoperative management.
Elderly patients account for nearly half of the surgical practice for ICC. 28The poor general condition of elderly patients naturally causes a preconceived assumption that elderly patients cannot tolerate complex or difficult procedures. 30As the mainstay procedure to treat primary liver malignancies, the safety of hepatectomy in elderly patients has aroused controversy.In a large cohort with 27,094 candidates, the rate of mortality in patients who underwent hepatectomy increased notably with age but subsequently reached a plateau in septuagenarians. 4,31 y considering aging risk and selecting appropriate procedures, good outcomes after complex surgery could be attained even in septuagenarians. 31Due to the high prevalence and extremely negative prognostic effect of LNM, LND is currently widely recommended as a standardized procedure in curative surgery for ICC. 32Nevertheless, a meta-analysis indicated that LND markedly increased the risk of morbidity with an elevated incidence of postoperative complications and brought no obvious survival benefit. 16Therefore, the clinical value of LND is basically considered a procedure to accurately stage and guide adjuvant therapy decisions. 33Consistently, 34 the current study indicated that patients with unstaged nodal status showed superior OS to those with LNM but inferior OS to those with negative nodal status (32 vs. 20 vs. 55 months, P = 0.001, Fig. 5), likely alluding to the underestimation of staging of a subset of patients without LND.Patients with unstaged nodal status are actually a combination of negative and understaged positive nodal patients.Indeed, the incidence of occult LNM in ICC patients without preoperative evidence of LNM was reported to be as high as 32.3-40.6%. 18,29 f note, TNLE affects the odds of identifying LNM (Table 2).Consistent with AJCC recommendations, 35 harvesting at least 6 LNs had the greatest discriminatory ability to identify LNM (Fig. 2).Accurate staging of nodal status contributes to guiding postoperative management, considering that patients with LNM are likely to benefit from adjuvant chemotherapy (Fig. 6A).The results were consistent with Qiao et al.'s study. 18he performance of LND for elderly ICC patients should involve considering not only the accuracy of staging information but also the risk of perioperative mortality and the benefit of long-term survival.In this nationwide cohort of elderly ICC patients aged ≥ 70 years, the total perioperative mortality was 2.6%, which is similar to that in Hiroko et al.'s study. 31Neither before nor after PSM was a significant difference in perioperative mortality observed between the LND group and the nLND group.Similarly, Zorays et al. reported that the perioperative mortality of ICC patients who underwent simultaneous LND or not was 5.7% and 5.9%,   respectively. 28In particular, both the total adverse events and life-threatening complications related to LND were comparable among the LND and nLND groups. 18,36 s such, the risk of implementing LND for elderly ICC patients is manageable.
Regarding long-term survival, the median OS in the LND group was significantly longer than that in the nLND group (44 vs. 32 months, P = 0.045; Fig. 3B), which was contrary to the aforementioned meta-analysis. 16Notably, the studies included in the meta-analysis were all retrospective and mostly single-center, with great heterogeneity in population characteristics. 13Namely, comparisons between patients who did and did not receive LND were often unmatched.In general, patients who received LND were at high risk for LNM with a poorer prognosis. 28Unbalanced population characteristics could probably minimize the survival differences of the two populations. 18,37 n the current well-matched study, elderly patients who underwent LND showed superior OS to those who did not, and LND was demonstrated to be an independent protective factor for Fig. 5 Overall survival of patients with negative, positive, or unstaged nodal status OS (Table 3).By further subgroup analysis, we found that elderly patients with relatively low tumor-burden characteristics and well-preserved liver function were more likely to benefit from LND (Fig. 4).In patients with the above features, the probability of systemic spreading 29 and postoperative complications 24 is lower.Thus, LND can be employed safely to remove lesions confined to the liver and regional lymph nodes.Notably, patients with extremely advanced age (> 80 years) might not benefit from LND. Older patients (> 80 years) are highly heterogeneous, are in poor general condition, have uncertain efficacy for surgery, and are rarely included in randomized clinical trials. 38In addition, the average life expectancy for Americans in 2020 will be approximately 77 years. 39For patients aged > 80 years, the potential benefit of LND for improving prognosis is probably discounted by the natural aging process itself.
To further investigate the significance of TNLE, unexpectedly, we found that AD-LND could not improve OS (mOS, 41 vs. 47 months, P = 0.612) but increased the odds of positive node retrieval by 6.49-fold (95% CI 1.98-21.28;P = 0.002) with respect to inAD-LND (Table 2).Thus, the prognostication of LND may be partly attributed to the definite evaluation of nodal status.Conversely, patients with AD-LND showed a longer median OS than inAD-LND patients in the LNM group (56 vs. 9 months, P = 0.043, supplementary 1).Similarly, Sahara et al. evaluated the therapeutic index of TNLE > 6 in node-positive patients with a value of 17.8 and indicated prolonged cancer-specific survival in node-positive patients with ≥ 3 LNs examined, which seems to confirm a survival benefit provided by LND. 19In contrast, several studies demonstrated the presence of LNM symbolizing a systemic disease in which the survival benefit of surgical treatment was limited. 40The exact prognostic value and appropriate candidates for LND need to be further confirmed in well-designed prospective multicenter studies.
This study has some limitations.First, owing to its retrospective nature, some biases could not be completely excluded between the two groups, even though a wellmatched PSM was conducted.In this national cohort, perioperative management and surgical procedures, including the extent of LND and postoperative chemotherapy regimen, varied from each center, and detailed data on surgical margins, and tumor biomarker levels were not recorded.The results should be interpreted with caution given the missing clinical information.Even though adequate dissection was defined as harvesting at least 6 LNs in the current study, it may not exactly correspond to LND in the ideal sense.In the future, prospective well-designed multicenter trials should be conducted to further confirm the clinical benefit of LND we observed.
In conclusion, advanced age (≥ 70 years) was not a contraindication for LND, which facilitates accurate nodal staging and guides postoperative management.Appropriately selected elderly populations could benefit from LND without increasing perioperative mortality.Notably, the purpose of the procedure should be weighed against its limited prognostic benefit when LND is performed in extremely elderly patients (> 80 years).

Fig. 1
Fig. 1 Flow chart of patient enrollment.One patient died perioperatively in the adequate-lymphadenectomy (AD-LND) group, which was excluded from the survival analysis

Fig. 3
Fig. 3 Overall survival of patients receiving lymphadenectomy (LND) or not before (A) and after propensity score matching (B)

Fig. 4
Fig. 4 Forest plot of subgroup analysis stratified by risk factors in the matched cohort.nLND, non-lymphadenectomy; LND, lymphadenectomy

Fig. 6
Fig. 6 Overall survival of patients who received postoperative chemotherapy (A-C) or radiotherapy according to nodal status (D-F)

Table 1
Clinicopathological characteristics before and after PSM Values are presented as medians (IQRs) for quantitative variables and n (%) for categorical variables Abbreviations: PSM, propensity score matching; nLND, non-lymphadenectomy; LND, lymphadenectomy; SMD, standardized mean differences; IQR, interquartile range; TNLE, total number of lymph nodes examined

Table 3
Univariate and multivariate analysis of OS before and after PSM Abbreviations: OS, overall survival; PSM, propensity score matching