Objective Endoscopic ultrasound-guided fine needle aspiration (EUS-FNA) allows preoperative tissue confirmation of malignancy, but fear of tumour cell dissemination along the needle track has limited its use. We hypothesised that if tumour cell dissemination occurs with EUS-FNA, survival after complete resection would be impaired. We aimed to evaluate the association of preoperative EUS-FNA with long-term outcomes of patients with resected pancreatic cancer.
Design Using the linked Surveillance, Epidemiology, and End Results (SEER)–Medicare data, we identified patients with locoregional pancreatic cancer who underwent curative intent surgery from 1998 to 2009. The patients who received EUS-FNA within the peridiagnostic period were included in the EUS-FNA group. Patients who did not receive EUS evaluation or who underwent EUS without FNA were included in the non-EUS-FNA group. Overall survival and pancreatic cancer-specific survival were compared after controlling for relevant covariates.
Results A total of 2034 patients with pancreatic cancer were included (90% pancreatic adenocarcinoma). Of these, 498 (24%) patients were in EUS-FNA group. Patients with multiple comorbidities and more recent diagnosis were more likely to receive EUS-FNA. In multivariate analysis, after controlling for age, race, gender, tumour histology, tumour stage, tumour grade, tumour location, SEER site, year of diagnosis, undergoing percutaneous aspiration/biopsy, Charlson Comorbidity Index, radiation and chemotherapy, EUS-FNA was marginally associated with improved overall survival (HR 0.84, 95% CI 0.72 to 0.99), but did not affect cancer-specific survival (HR 0.87, 95% CI 0.74 to 1.03).
Conclusions Preoperative EUS-FNA was not associated with increased risk of mortality. These data suggest that EUS-FNA can be safely performed for the work-up of suspicious pancreatic lesions.
- ENDOSCOPIC ULTRASONOGRAPHY
- PANCREATIC CANCER
Statistics from Altmetric.com
Significance of this study
What is already known on this subject?
Although endoscopic ultrasound-guided fine needle aspiration (EUS-FNA) provides valuable diagnostic information, there have been concerns that preoperative EUS-FNA may adversely affect patients with pancreatic cancer because of the potential for tumour seeding or dissemination of tumour cells via a puncture into the peritoneal cavity.
Many physicians still avoid EUS-FNA out of fear of tumour seeding, particularly in patients with potentially resectable tumours.
What are the new findings?
The current study used the Surveillance, Epidemiology and End Results (SEER)–Medicare-linked database to identify surgically resected locoregional pancreatic cancers and compared long-term outcomes with regard to those who did or did not have EUS-FNA while controlled for other factors.
We found that preoperative EUS-FNA did not adversely affect overall survival or cancer-specific survival.
How might it impact on clinical practice in the foreseeable future?
This information should reassure physicians that EUS-FNA can be safely performed for the work-up of pancreatic lesions when there is a suspicion of pancreatic cancer.
Endoscopic ultrasound-guided fine needle aspiration (EUS-FNA) is a commonly used method for definitive tissue diagnosis of suspicious pancreatic lesions, with an overall sensitivity and specificity of 85% and 98%, respectively.1 ,2 This procedure is generally safe with an overall rate of EUS-FNA-related morbidity of 0.98% and mortality of 0.02%. Most associated adverse events are acute pancreatitis and postprocedure pain, which are mild to moderate in severity.3
Gastric wall implantation and dissemination of tumour cells via a puncture into the peritoneal cavity by EUS-FNA is a rare but serious adverse event of this procedure. Malignant gastric wall seeding after EUS-FNA has been reported in the literature.4–6 As a consequence, it has been suggested that EUS-FNA should be avoided in patients with potentially resectable tumours, particularly if the surgical resection will not include the needle tract, such as lesions in the body/tail of pancreas. Despite this theoretical limitation, EUS-FNA is frequently performed in patients with pancreatic lesions.
There have been a few clinical studies conducted to address this question. The reports from a single tertiary care centre comparing patients who received preoperative EUS-FNA with those who did not receive EUS-FNA suggested that EUS-FNA was not associated with an increased rate of gastric or peritoneal cancer recurrence in patients with pancreatic cancer treated with surgical resection.7 ,8 The limitation of these studies is that they included a relatively small sample size that may be underpowered to determine small differences in long-term outcomes between the two groups. To date, no large population-based study has assessed whether EUS-FNA has a negative impact on outcomes of patients with resected pancreatic cancer.
We hypothesised that if tumour cell dissemination occurs with EUS-FNA, survival after complete resection would be impaired. In this study, we analyse the Surveillance, Epidemiology and End Results (SEER) database linked to Medicare claims to evaluate the association of preoperative EUS-FNA with outcomes of pancreatic cancer-specific survival and overall survival of patients with locoregional pancreatic cancer who were undergoing curative intent surgery compared with those who did not receive EUS-FNA.
Study population and data source
We identified all patients aged 66 years or older with first primary pancreatic cancer diagnosed between 1998 and 2009 from the linked SEER-Medicare database for this analysis. We included (1) patients with International Classification of Diseases for Oncology (ICD-Third edition) histology codes consistent with adenocarcinoma (8140, 8141, 8143, 8211 and 8480–8481), mucinous cystadenocarcinoma (8470, 8471), intraductal papillary mucinous carcinoma of pancreas (8453), malignant endocarcinomas (8150–8157) and malignant carcinoid tumours (8240–8249); (2) patients with localised or regional (locoregional) pancreatic cancer based on SEER historic stage; (3) patients who underwent curative intent surgery and (4) patients who were enrolled in both Medicare parts A and B without a health maintenance organisation for 12 months before their cancer diagnosis.
The following patients were excluded from the analysis: patients with cancer diagnosed on autopsy or death certificate only, patients with no Medicare part A or B coverage from 1 year before (because outpatient claims were required), patients who were enrolled in a Medicare Health Maintenance and patients who were eligible for Medicare on the basis of end-stage renal disease or disability.
We extracted the relevant demographic data, tumour characteristics and survival through the end of 2010. The Deyo adaptation of the Charlson Comorbidity Index was used to assess comorbidities present in the study population.9
All Outpatient Standard Analytic Files and National Claims History claims during the period of 1 month before and 3 months after the date of pathological diagnosis were examined for procedure codes for EUS (CPT-4 43231, 43259), EUS radiological interpretation (CPT-4 76975) and EUS/FNA (CPT-4 43232, 43242). Using a similar approach, we also identified patients who underwent ‘percutaneous aspiration or biopsy of pancreas or abdominal/retroperitoneal mass’ (ICD-9: 52.11; CPT: 10021, 10022, 48102, 49180).
Patients were defined as having curative intent surgery if SEER data or Medicare claims indicated partial or total pancreatectomy, radical pancreaticoduodenectomy or excision of a lesion of the pancreas. To identify chemotherapy, the Medicare claims were reviewed to determine whether there were any general codes or diagnoses for chemotherapy administration or claims for specific agents (fluorouracil and gemcitabine) within 6 months after the time of cancer diagnosis, as reported by the Patient Entitlement and Diagnosis Summary File.10 Radiation therapy was ascertained from SEER data and Medicare claims for the 9-month period after diagnosis.11 We identified patients who subsequently had postprocedural complications (pancreatitis within 1 day; postprocedure haemorrhage, perforation and infection within 30 days) after EUS procedure and before pancreatic surgery. All Medicare claims with procedure codes or revenue centre codes were obtained from Medicare Provider Analysis and Review, Outpatient Claims files, physician/supplier files and Durable Medical Equipment. The list of treatment codes is included in the online supplementary appendix A.
The study population was divided into two groups: patients who received EUS-FNA (EUS-FNA group) and patients who did not receive EUS, or who underwent EUS but not FNA (non-EUS-FNA group). Baseline characteristics were compared between the groups with χ2 test or independent t test, as appropriate. Trends in ordinal data were evaluated using the linear-by-linear association test. The linear-by-linear test of trend offers a measure of significance for ordinal variables (calendar year ordered from lowest to highest). A logistic regression model was performed to study the factors associated with receiving EUS-FNA.
Survival time was calculated from the date of diagnosis to the date of death or the date of the last follow-up. Overall survival and pancreatic cancer-specific survival were determined from SEER cause of death data. For analysis of pancreatic cancer-specific survival, deaths from causes other than pancreatic cancer were treated as censored observations. We calculated Kaplan–Meier estimates for overall and cancer-specific survival. Multivariate analyses were performed using the Cox proportional hazard model to identify factors associated with cancer-specific survival and overall survival after adjustment for other relevant factors, including age (continuous variable), sex, race, stage of cancer, Charlson Comorbidity Index, tumour location, histological pathology, tumour grade, SEER regions (Northeast, Midwest, South and West, as defined by the US Census Bureau), year of diagnosis, undergoing percutaneous aspiration/biopsy, chemotherapy and radiation therapy. We included all relevant patient demographics and tumour variables in the multivariable analysis. Adjusted HRs and 95% CIs were reported. The significance level for all tests was two-sided at 5%. The proportional hazard assumption was assessed by visually examining the log-minus-log survival plots for each variable. The Cox regression models were used if the proportional hazards assumptions were not violated (ie, when the log-minus-log curves did not cross). An examination of the log-minus-log plots demonstrated that the proportional hazards assumption was met. All data were analysed using SAS (SAS Institute, Cary, North Carolina, USA) and SPSS Statistics software (V.16; IBM Corporation, USA). For the main analysis, we compared the long-term outcomes of all patients with pancreatic cancer. We next performed a subgroup analysis of patients with pancreatic adenocarcinoma and analysis according to the location of cancer in the pancreas (head and body/tail).
A small proportion of patients also underwent percutaneous aspiration or biopsy of pancreas or abdominal/retroperitoneal mass before their surgical resection. Because previous study has showed that the risk of peritoneal seeding increased with percutaneous samplings when compared with EUS-FNA,12 we performed a sensitivity analysis to assess the robustness of our findings by repeating our analyses after excluding patients who underwent percutaneous tissue samplings.
The study was approved by the Mayo Clinic Institutional Review Board.
A total of 2034 patients with locoregional pancreatic cancer fulfilled the inclusion and exclusion criteria. This included 498 patients (24%) in the EUS-FNA group and 1536 patients (76%) in the non-EUS-FNA group. In the EUS-FNA group, 457 patients (92%) received one EUS-FNA procedure and 40 patients (8%) underwent two EUS-FNA procedures. In the non-EUS-FNA group, 1339 patients (87%) did not receive EUS evaluation and 197 patients (13%) underwent EUS without FNA. The distribution based on stage of disease was localised stage in 412 patients (20%) and regional stage in 1622 patients (80%). The majority of cases had pancreatic adenocarcinoma (n=1839, 90%). Of all patients, 378 patients (18.6%) also underwent percutaneous aspiration or biopsy of pancreas or abdominal/retroperitoneal mass. Baseline characteristics of patients and cancer-related variables in both groups are shown in table 1. Patients in the EUS-FNA group had higher comorbidity scores than patients in the non-EUS-FNA group. The frequency of EUS-FNA increased during the course of the study and was more likely to be used for patients who resided in the West and less likely for patients in the South. Percutaneous aspiration/biopsy and chemotherapy were performed more frequently in the EUS-FNA group.
Of the 498 patients who underwent EUS-FNA, 11 (2.2%) developed pancreatitis requiring hospitalisation. None of the patients in the EUS-FNA group encountered postprocedure haemorrhage, perforation or infection.
Overall use of EUS-FNA increased during the time periods examined, with 10% (18 of 172 patients) in 2001 to 47.1% (96 of 204 patients) in 2009; p<0.001 (figure 1). In logistic regression analysis, the OR for EUS-FNA in patients diagnosed between 2006 and 2009 compared with those diagnosed between 1998 and 2001 was 17.21 (95% CI 10.17 to 29.10). Other significant factors associated with receiving EUS-FNA included higher comorbidity scores (OR 1.37, 95% CI 1.08 to 1.72), patients who resided in the West (OR 1.89, 95% CI 1.39 to 2.55) and Midwest (OR 1.76, 95% CI 1.19 to 2.63) as compared with those who resided in the South (table 2).
During a mean follow-up time of 21 months (range 0–140 months), 285 patients (57%) in the EUS-FNA group and 1167 patients (76%) in the non-EUS-FNA group died. Pancreatic cancer was identified as the cause of death for 251 patients (50%) in the EUS-FNA group and for 980 patients (64%) in the non-EUS-FNA group.
Median overall survival in the EUS-FNA and non-EUS-FNA groups, estimated by the Kaplan–Meier method, was 22 months (95% CI 18.8 to 25.2) and 15 months (95% CI 13.9 to 16.1), respectively. In the multivariate Cox regression model after adjusting for other variables, receipt of EUS-FNA had a borderline significant association with improved overall survival (HR 0.84, 95% CI 0.72 to 0.99, p=0.03). The significant variables associated with decreased overall survival included older age at diagnosis, non-Caucasian race, increasing stage of disease, higher Charlson Comorbidity Score, adenocarcinoma-type cancer, higher tumour histological grade, earlier year of diagnosis, not receiving chemotherapy or radiation therapy (table 3).
In an analysis combining patients who received preoperative EUS (both with and without FNA), the multivariate analysis after adjusting for age, sex, race, stage of cancer, Charlson Comorbidity Index, tumour location, histological pathology, tumour grade, SEER regions, year of diagnosis, undergoing percutaneous aspiration/biopsy, chemotherapy and radiation therapy, undergoing EUS (both with and without FNA) was associated with improved overall survival (HR 0.77, 95% CI 0.67 to 0.87, p<0.001).
Median cancer-specific survival in the EUS-FNA and non-EUS-FNA groups was 24 months (95% CI 20.6 to 27.3) and 18 months (95% CI 16.5 to 19.4), respectively. In the multivariate Cox regression model after adjusting for other variables, HR for cancer-specific survival in the EUS-FNA group was also not different from that of the non-EUS-FNA group (HR 0.87, 95% CI 0.74 to 1.03, p=0.12). Factors associated with decreased cancer-specific survival were older age, non-Caucasian race, increasing stage of disease, adenocarcinoma-type cancer, higher tumour histological grade, earlier year of diagnosis and not receiving radiation therapy (table 4; see online supplementary figure S1).
In a univariate analysis, overall survival and cancer-specific survival of patients who underwent percutaneous puncture/aspiration were not different from those who did not; median overall survival was 17 months (95% CI 14 to 19) and 17 months (95% CI 15 to 18; p=0.88) and median cancer-specific survival was 19 months (95% CI 15 to 22) and 20 months (95% CI 18 to 21; p=0.21), respectively. In a multivariable analysis, undergoing percutaneous aspiration or biopsy was not associated with impaired overall survival or cancer-specific survival (HR 1.06, 95% CI 0.92 to 1.24 and HR 1.12, 95% CI 0.96 to 1.31, respectively).
In a subgroup analysis of 1839 patients with pancreatic adenocarcinoma, median overall survival was 21 months (95% CI 18 to 23) in the EUS-FNA group and 15 months (95% CI 14 to 16) in the non-EUS-FNA group. In the multivariate Cox regression model, HR for overall survival in the EUS-FNA group was not different between the groups (HR 0.87, 95% CI 0.75 to 1.02, p=0.09). Median cancer-specific survival of patients with pancreatic adenocarcinoma was 23 months (95% CI 20 to 25) in the EUS-FNA group and 16 months (95% CI 15 to 17) in the non-EUS-FNA group. In the multivariate Cox regression model, there was no difference in HR for pancreatic cancer-specific survival (HR 0.91, 95% CI 0.77 to 1.07, p=0.26).
Next, we explored whether patients undergoing transgastric FNA had poorer outcomes when compared with those with transduodenal FNA, where the needle path is typically resected. In the multivariate Cox regression analysis according to the location of cancer in the pancreas (head and body/tail), there was no difference in overall or cancer-specific survival (table 5).
In the analysis after excluding patients who had undergone percutaneous aspiration or biopsy of pancreas or abdominal/retroperitoneal mass, HR of the EUS-FNA group for overall survival and cancer-specific survival was not different from that of the non-EUS-FNA group (overall survival: HR 0.84, 95% CI 0.70 to 1.01, p=0.07 and cancer-specific survival: HR 0.86, 95% CI 0.70 to 1.04, p=0.12).
EUS-FNA is a well-established technique for tissue diagnosis of suspicious pancreatic lesions. The main objective of this procedure is to accurately identify malignant lesions and spare the patients with benign condition and unnecessary surgical resections. Although EUS-FNA of pancreatic lesions is generally regarded as a safe procedure, there have been concerns of tumour cell seeding along the needle track or within the peritoneum caused by EUS-FNA. In this study, we used the population-based SEER-Medicare-linked database to identify a large cohort of patients with locoregional pancreatic cancer who underwent curative intent surgery. We found that undergoing EUS-FNA was not associated with increased overall or cancer-specific mortality. These data suggest that EUS-FNA can be safely performed for the work-up of suspicious pancreatic lesions, even for the patients with potentially resectable cancers.
In fact, we found that receipt of EUS-FNA was marginally associated with improved overall survival in the multivariate analysis (HR 0.84, 95% CI 0.72 to 0.99). When we compared patients who received EUS both with and without FNA and those who did not receive EUS in multivariate analysis, receipt of EUS was independently associated with improved overall survival. This finding was in line with a previous population-based study that showed that EUS evaluation was an independent predictor of improved survival in patients with locoregional pancreatic cancer, most probably as a result of improved stage-appropriate management, including more selective performance of curative intent surgery and perioperative adjuvant therapy.13
Several cases of tumour seeding along a needle tract in pancreatic adenocarcinoma occurring at the gastric wall following EUS-FNA of pancreatic body or tail have been reported in the literature,4–6 but this has never been reported for EUS-FNA of pancreatic head lesions. The theoretical explanation is that tumours in the pancreatic body/tail are generally sampled via a transgastric approach and the path of the needle is not resected at the time of subsequent pancreatectomy, whereas in the transduodenal approach, the site of needle tract in EUS-FNA for pancreatic head lesions (the duodenal bulb) is included within the area of surgical resection. In the analysis, according to the tumour location in the present study, we also found that preoperative EUS-FNA was not associated with poorer long-term outcomes, whether it was performed for lesions in the head or body/tail.
The concerns over cancer cell seeding along the needle track or within the peritoneum caused by EUS-FNA of pancreatic lesions have arisen from numerous case reports.4 ,5 ,6 ,14 ,15 In a retrospective study of 175 patients with intraductal papillary mucinous neoplasm, EUS-FNA did not increase the frequency of postoperative peritoneal seeding.16 Because a majority of patients included in this study had benign intraductal papillary mucinous neoplasm, the results could not be applicable to patients with malignant solid tumours. A study by Ikezawa et al17 reported that patients with pancreatic adenocarcinoma who had undergone EUS-FNA (n=56) did not significantly increase the risk of peritoneal carcinomatosis. The impact of EUS-FNA on survival was not reported in this study.
There have been a few comparative studies to assess the risk of preoperative EUS-FNA on pancreatic cancer recurrence.7 ,8 In a single-centre study of 256 patients with resectable pancreatic cancer,7 the frequency of gastric/peritoneal recurrence in patients who underwent preoperative EUS-FNA was not different from those who did not receive EUS-FNA (13 of 168 patients (7.7%) vs 6 of 39 patients (15%), respectively, p=0.21). Beane et al8 evaluated postoperative long-term outcomes of 63 patients with pancreatic adenocarcinoma undergoing distal pancreatectomy. The patients who received preoperative EUS-FNA, as compared with those who did not, had similar recurrence-free or overall survival. This study, however, did not address whether EUS-FNA of pancreatic head lesion, which occur in the majority of patients with pancreatic cancer, affects the long-term outcomes.
The strength of our study is the large number of patients included, which was powered to find relatively small differences in survival between the groups and allows the more precise estimation of survival rates of patients. The SEER-Medicare registry contains population-based data, and therefore, the results are more generalisable to the US population than data from tertiary cancer centres. In addition, Medicare claims data allow us to control for comorbidities conditions and receipt of chemoradiation, which are the potential confounders.
There are several limitations in our study. First, we based our determination of EUS-FNA from claims data. The accuracy of procedural coding of GI procedures in Medicare claims has been studied only on a limited basis. However, where studied, the sensitivity and positive predictive value of coding of procedural claims were high.18 No data regarding surgical margin status, the type or size of the needle used for FNA, number of needle passes and location of needle passage is provided in the SEER-Medicare database. Therefore, we could not evaluate the risk of cancer dissemination according to these factors. Due to the use of Medicare-eligible patients, the study was limited to a population aged <65 years. The results of this study may not be generalisable to younger patients with pancreatic cancer. However, it should be noted that the majority of patients with pancreatic cancer are in this elderly age group.19 We did not report the rate of cancer recurrence in our study because neither SEER nor Medicare provides data on disease recurrence. However, SEER provides cause of death to calculate cancer-specific survival, which is temporally correlated with cancer recurrence. Finally, this is a non-randomised study. Our results could be influenced by selection bias and confounders. We observed differences in the baseline characteristics of patients who had and did not have EUS-FNA. Although we controlled for known confounding variables, our results might still be influenced by unmeasured and residual confounders such as endoscopists’ experience and physician's practice pattern. There is a potential that centres that use advanced technologies, such as EUS-FNA, also have more oncological expertise and better outcomes, although this is hard to prove or disprove. Despite this limitation, a large randomised trial of patients to answer this question is unlikely to be feasible. Thus, in the absence of information from a randomised, controlled trial, data from a large population-based cohort are probably the best source of data to answer this question.
In summary, this study shows that preoperative EUS-FNA is not associated with increased risk of cancer-specific or overall mortality in the patients with resected pancreatic cancer. Gastric recurrence or peritoneal seeding of cancer cells from EUS-FNA does not appear to have clinical significance or affect long-term outcomes in these patients. This information reassures physicians that tissue diagnosis by EUS-FNA is safe for the work-up of suspicious pancreatic lesions. Randomised controlled trials of staging with and without EUS-FNA are needed to definitively prove if it is valuable.
We would like to thank Alison Dowdell for assistance with manuscript preparation.
Contributors All authors have approved the final draft of the manuscript. SN: Conception and design, analysis and interpretation of data, drafting of the article. KW: Conception and design, analysis and interpretation of data, critical revision of the article for intellectual content. NDS: Conception and design, analysis and interpretation of data. MBW: Conception and design, critical revision of the article for intellectual content.
Competing interests MBW receives research funding (unrelated to this manuscript) from COSMO pharmaceuticals, Olympus, Boston Scientific, Ninepoint and US Endoscopy.
Ethics approval Mayo Clinic Institutional Review Board.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.