Association between biopsy method and development of peritoneal metastases in perihilar cholangiocarcinoma

Abstract

Background:

In patients with perihilar cholangiocarcinoma (PHC), there is concern that transperitoneal (TP) biopsy may seed tumor in the peritoneal cavity, increasing risk of peritoneal metastases (PM).

Methods:

A retrospective review of patients undergoing surgery for PHC (1991–2014) was performed. Clinicopathologic characteristics and incidence of PM at the time of index surgery, and one and two years after surgery were compared in patients who did vs. did not undergo TP biopsy.

Results:

Among 262 patients who underwent surgery, 37 had undergone TP biopsy, and 225 had undergone intraluminal biopsy or had no biopsy. No differences in demographic or clinicopathologic characteristics were noted between groups. The incidence of PM at surgery was not significantly different between TP and non-TP biopsy patients (5.4% vs. 7.6%, p > 0.9). Among 243 patients who did not have PM at surgery, the cumulative incidence of PM in the TP and non-TP biopsy groups were not different at one year (11.4% [95%CI 3.5–24.4] vs. 10.8% [95%CI 7.0–15.5]) or two years (20.3% [95%CI 8.7–35.2] vs. 20.1% [95%CI 14.9– 25.9]) (p = 0.7).

Discussion:

Although PM commonly occurs in patients with PHC, TP biopsy was not associated with higher incidence of PM at surgery or at one or two years after surgery.

Introduction

Perihilar cholangiocarcinoma (PHC) is a rare malignancy with a poor prognosis. The age-adjusted incidence of PHC is 0.38 per 100,000 in the United States, with an overall 5-year survival of 9.1%.¹ Complete resection is the only curative option and is associated with a 5-year survival of 30–40%.^2,3 Although rarely utilized, 5-year survival after transplant is as high as 64%.⁴ Unfortunately, only 27–35% of patients are eligible for resection or transplant at the time of presentation, often due to metastatic disease.^1,5 The peritoneum is a frequent site of metastatic disease, with reports of peritoneal metastases (PM) in 10–20% of patients at the time of diagnosis.⁶ Concern has been raised that PHC-related interventions that violate the peritoneum, including some biopsy procedures that traverse the peritoneum, may increase a patient’s risk of developing peritoneal dissemination.⁷ However, the extent to which transperitoneal (TP) biopsy increases this risk is not well-studied.

Peritoneum-violating interventions are often considered in PHC both for biliary drainage and biopsy and may cause tumor seeding and disease dissemination.⁸ Early studies of percutaneous transhepatic biliary drainage PTBD in PHC suggested an increased risk of PM,⁹ but in larger studies, PTBD was shown not to be associated with higher rates than other forms of drainage.^10,11 TP biopsies are also considered in PHC because of the difficulty diagnosing early disease on imaging and the low sensitivity of endoscopic brush biopsies.^12,13 One small study of patients planned for liver transplantation for PHC found that 5/6 (83%) patients who had positive preoperative TP biopsies had PM at staging laparoscopy. This was substantially higher than- the 14/175 (8%) patients who did not have TP biopsies.⁷ No larger studies have attempted to confirm these findings. However, these findings have led to the exclusion of any patient who had any TP biopsy from receiving liver transplantation at some institutions.¹⁴

Associations between TP biopsies and the development of PM in PHC are challenging to evaluate because of the rarity of the disease and the resultant difficulties establishing causation. The aim of this study was to further analyze the relationship between TP biopsy and PM in a large cohort of patients with PHC undergoing surgery.

Methods

Patient population

The project was approved by the Institutional Review Board at Memorial Sloan Kettering Cancer Center (MSKCC). Consecutive patients who underwent exploratory laparotomy or diagnostic laparoscopy for PHC at MSKCC from 1991 to 2014 were identified from a prospectively collected departmental database. Patients with multiple active cancers, no documentation of preoperative biopsy status, or less than six months of follow-up were excluded from the analysis.

Patient characteristics, clinical management, including preoperative biopsy type and timing, and pathologic details were obtained from the electronic medical record. Image-guided percutaneous and endoscopic-ultrasound guided transduodenal biopsies were classified as TP biopsies. Bile duct brushings were classified as intraluminal (IL) biopsies. Patients who had both TP biopsy and IL biopsy were classified as TP biopsies. Patients were classified as “no biopsy” if the absence of biopsy was explicitly documented in the medical record. The specific indications for performing the biopsy were not accounted for given that patients often come to our institution for further management after they have undergone biopsies at other institutions. Many of the biopsies were done at the discretion of treating physicians at outside institutions. Presence of PM at the time of initial surgery was determined from review of operative notes and pathology reports. The margin of resection was determined from operative and pathology reports and defined as R0 for microscopically negative margins with no tumor present in the resection bed, R1 as grossly negative margins with microscopic tumor present at the resection margins, and R2 as gross tumor present after surgery. Patients who had R2 resections or underwent surgery for palliative bypass procedures were classified as unresectable. Presence of PM after surgery was determined from review of operative notes from subsequent abdominal surgical interventions or from cross-sectional imaging and/or biopsies documenting clear evidence of PM.

Statistical analysis

Patient characteristics, clinical management, disease pathology, and PM at the time of index operation and after surgery were compared between patients who underwent preoperative TP biopsies and those who did not. Cumulative incidence of PM after surgery, regardless of other sites of metastases, was also compared between patients according to biopsy type and biopsy positivity.

Patient demographics and clinical and treatment characteristics were summarized using frequency and percentages for categorical covariates, and median and range for continuous covariates, and compared using Fisher’s exact test and the Wilcoxon rank-sum test. Associations between the type of biopsy and occurrence of PM at the time of surgery were analyzed using a Fisher’s exact test. A multivariable model was not considered for this endpoint since we only observed 19 patients with PM at the time of surgery.

Cumulative incidence of PM from the time of laparotomy was estimated using competing risks methods in order to correctly estimate marginal probability of PM in the presence of competing events, such as death, and compared between subgroups using Gray’s test.^15,16 Patients who already had the event of PM at the time of surgery (n = 19) were excluded from this analysis.

Overall survival (OS) was calculated for all patients from the time of laparotomy and estimated using Kaplan– Meier methodology. All statistical analyses were performed using R version 3.6.0 (R Foundation for Statistical Computing, Vienna, Austria). All P-values were two-sided; p < 0.05 was considered significant.

Results

Patient population

A total of 263 patients underwent surgery for PHC at MSKCC from January 1991 to December 2014. One patient was excluded from the analysis because of missing biopsy data. Among these 262 patients, mean age was 67 years (range 53– 73), 43% were male, and mean BMI was 25.5 kg/m². Thirty-seven (14.1%) underwent a preoperative TP biopsy, 134 (51.2%) underwent IL sampling alone, and 91 (34.7%) had no tissue sampling performed.

Between patients who had TP biopsy performed and those who did not (i.e., TP biopsy vs. IL biopsy/no biopsy), there was no difference in demographic features, including age, sex, BMI, smoking history, or alcohol abuse. Additionally, there were no differences in preoperative treatment, receipt of neoadjuvant chemotherapy or radiotherapy, or margin of resection. Slightly more patients who underwent TP biopsy also had pre-operative PTBD compared to those that did not undergo TP biopsy (40% vs 34%, p = 0.042; Table 1). The median time from diagnosis to surgery for the whole cohort was 38 days and did not significantly differ between patients who underwent TP biopsy and those who did not (40 vs. 38 days; p = 0.6). Similarly, among those who had any biopsy, the median time from biopsy to surgery was 38 days and did not significantly differ between patients who underwent TP biopsy and those who underwent only IL biopsy (32 vs. 38 days; p = 0.2). Finally, among patients who underwent resection (n = 155), tumor size, presence of positive lymph nodes, tumor differentiation, and presence of papillary histology did not differ between those that had TP biopsy vs. those that did not. There was no significant difference in the proportion of patients who had R0 resection, R1 resection or unresectable between those that had TP biopsy vs. those that did not (Table 2).

Table 1.

Demographic and clinical data the cohort (N = 262) by biopsy type

	No TP Biopsya	TP Biopsy	p-valuec
	n =225b	n = 37b
Epidemiology
Age, years	67 (57– 73)	63 (53– 70)	0.2
Sex			0.3
Male	93 (41%)	19 (51)
Female	132 (59%)	18 (49%)
BMI	25.5 (23.2– 28.7)	24.3 (20.9– 27.6)	0.06
Smoking History			0.8
Yes	84 (41%)d	12 (38%)d
History of Alcohol Abuse			0.7
Yes	21 (10%)d	2 (5%)d
Treatment
Neoadjuvant Therapy
Chemotherapy	7 (3%)	2 (5%)	0.6
Radiation	4 (2%)	2 (5%)	0.2
Preoperative Stent	173 (77%)	24 (65%)	0.2
None	46 (22%)	13 (37%)	0.042
Percutaneous transhepatic biliary drainage	70 (34%)	14 (40%)
Endoscopic biliary drainage	91 (44%)	8 (23%)
Unknown	18	2
Time from Diagnosis to Surgery, days	38 (21– 54)	40 (25– 62)	0.6
Time from Biopsy to Surgery, days	38 (24– 56)	32 (22– 50)	0.4
Margin Status			0.9
R0	84 (37%)	14 (38%)
R1	50 (22%)	7 (19%)
Unresectablee	91 (41%)	16 (43%)

BMI: body mass index; TP: transperitoneal.

^aNo TP Biopsy includes patients who had no pre-operative biopsy or pre-operative intraluminal biopsy only.

^bStatistics presented: median (IQR); n (%).

^cStatistical tests performed: Wilcoxon rank-sum test; chi-square test of independence; Fisher’s exact test.

^dFrequency reported excludes missing/unknown values.

^e“Unresectable” includes 4 patients who had R2 resections, patients who had peritoneal metastasis at the time of surgery, and patients who underwent palliative bypass procedures and therefore did not undergo resection.

Table 2.

Pathologic data among R0/R1 patients by biopsy type

	No TP Biopsy	TP Biopsy	p-valueb
	N= 134a	N= 21a
Pathology
Largest tumor Diameter, cm	2.50 (1.75– 3.55)	3.10 (1.85– 4.78)	0.2
Lymph Node			>0.9
Negative	98 (73%)	15 (71%)
Positive	36 (27%)	6 (29%)
Unknown	0	0
Perineural Invasion			0.4
No	42 (31%)	4 (20%)c
Yes	92 (69%)	16 (80%)c
Unknown	0	1
Lymphovascular Invasion			0.7
No	88 (67%)c	12 (60%)c
Yes	44 (33%)c	8 (40%)c
Unknown	2	1
Differentiation
Mod-Poor Differentiation Score	100 (78%)c	14 (74%)c	0.8
Unknown	5	2
Papillary Histology	25 (19%)	7 (33%)	0.15
Tumor Stage (AJCC 7th Edition)			0.4
0	5 (4%)	0 (0%)
1	12 (9%)	3 (14%)
2	62 (46%)	8 (38%)
3	36 (27%)	4 (19%)
4	19 (14%)	6 (29%)

TP: transperitoneal.

^aStatistics presented: median (IQR); n (%).

^bStatistical tests performed: Wilcoxon rank-sum test; chi-square test of independence; Fisher’s exact test.

^cFrequency excludes unknown/missing values.

Incidence of PM at surgery

Overall, at the time of surgery, 19/262 (7.3%) patients in the cohort were found to have PM. There was no difference in presence of PM at surgery based on preoperative biopsy. PM was identified at surgery in 2/37 (5.4%) patients with preoperative TP biopsy vs. 17/225 (7.6%) patients without preoperative TP biopsy (p > 0.9).

Cumulative incidence of PM after surgery

Among the 243 patients who did not have PM at the time of surgery, 72 (29.6%) ultimately developed PM during follow-up, confirmed from biopsies from subsequent abdominal surgical interventions (n = 10) or from cross-sectional imaging (n = 62). The cumulative incidence of PM among these 243 patients (excluding those with PM at surgery) at one- and two-years following surgery were 10.9% (95%CI: 7.3– 15.2) and 20.1% (95%CI: 15.3 – 25.5), respectively. There was no difference between the cumulative incidence of PM according to biopsy type. Among the 35 patients who had TP biopsy and no PM at the time of surgery, the cumulative incidences of PM at one- and two- years following surgery were 11.4% (95%CI: 3.5– 24.4) and 20.3% (95%CI: 8.7– 35.2). Among 208 patients who had no TP biopsy, the cumulative incidences of PM at one and two years after surgery were 10.8% (95%CI: 7.0– 15.5) and 20.1% (95%CI: 14.9– 25.9), respectively (p = 0.7 vs. TP; Fig. 1).

Figure 1.

Cumulative incidence of peritoneal metastases by biopsy type. No biopsy vs. intraluminal biopsy vs. transperitoneal biopsy, p = 0.3; no biopsy or intraluminal biopsy vs. transperitoneal biopsy, p = 0.7

Finally, we investigated the association between PM incidence and biopsy result. Of the 35 patients who underwent TP biopsy with documented results and did not have PM at the time of surgery, 25 (71.4%) were positive for PHC. Of the 123 patients who underwent IL biopsy with documented results (n = 122) and did not have PM at the time of surgery, 82 (67.2%) were positive. Cumulative incidence of PM did not differ between patients with positive and negative TP biopsies, positive and negative IL biopsies, and no biopsy performed (p = 0.4; Fig. 2).

Figure 2.

Cumulative incidence of peritoneal metastases by biopsy type and result. TP+ = positive transperitoneal biopsy; TP− = negative transperitoneal biopsy; IL+ = positive intraluminal biopsy; IL− = negative intraluminal biopsy

Recurrence and overall survival after R0 resection

Among those who did not have PM at surgery, 98 ultimately underwent R0 resection. Among those who had R0 resection, 18(18.4%) had recurrence over a mean interval of 37 months, and 6(6.1%) had PM as the first site of recurrence (Table 3). A slightly lower rate of recurrence was seen in patients that underwent compared to those that did not undergo TP (14.3% vs 19.0%). However, the incidence of PM as the first site of recurrence did not differ substantially between patients who underwent TP biopsy vs. those that did not (7.1% vs. 6.0%). The median OS among those who underwent TP biopsy was 40 months compared to 48 months among those who did not undergo TP biopsy. The sample sizes were too small to make any formal comparisons or test for statistical significance.

Table 3.

Recurrence^a and overall survival among patients who underwent R0 resection^b

	No TP Biopsy	TP Biopsy	Total
	n = 84	n = 14	n = 98
Recurrence
Any recurrence	16 (19.0%)	2 (14.3%)	18 (18.4%)
PM as site of first recurrence	5 (6.0%)	1 (7.1%)	6 (6.1%)
Overall Survival, median in months (range)	48.9 (17– 248)	41 (7– 91)	57 (7– 248)

TP = transperitoneal biopsy; PM = peritoneal metastasis.

^aOver a mean interval of 37 months.

^bSample sizes too small for statistical or formal comparisons.

Discussion

The potential for preoperative TP biopsy to seed the peritoneum in patients with PHC is not well-studied. In this study, we reviewed patients with PHC who were taken to the operating room for planned resection and identified no association between TP biopsy and increased incidence of PM at the time of or after surgery. These findings suggest that TP biopsy does not result in a significantly increased risk of PM in this study cohort.

Prior studies of biopsy in patients with PHC have generated conflicting conclusions about biopsy-related tumor seeding.^17–23 A review of patients undergoing liver transplantation for hilar cholangiocarcinoma found that 5/6 (83%) patients with positive preoperative TP biopsies had PM at staging laparoscopy, compared to 14/175 (8%) of patients who did not have TP biopsies.⁷ Furthermore, in several reported cases, PM appeared to be along needle tract sites.²⁴ These patients were unable to undergo transplantation or resection. This dramatic result prompted many transplant centers to consider prior TP biopsy a contraindication to transplantation in patients with PHC.¹⁴ However, a review of 150 patients with cholangiocarcinoma at any location, of whom 61 underwent endoscopic ultrasound-guided fine needle aspiration biopsy, found no association between biopsy type and either overall survival or recurrence-free survival.²⁵ The lack of association with recurrence was unexpected, given the strong association suggested by the aforementioned Heimbach et al. study, although this study did not explicitly address peritoneal spread of tumor.

In the current study, we analyzed the association between biopsy and peritoneal dissemination of disease in a cohort of patients with PHC planned for surgery. There were no significant demographic, therapeutic, or pathologic differences identified between patients who underwent TP biopsy and those who did not. PM was noted in a small number of patients at the time of surgery, but there was no difference in the incidence between patients who underwent TP biopsy and those who did not. The time from biopsy to surgery varied considerably between patients, but the median time from biopsy to surgery was less than two months, suggesting that in the short-term TP biopsy did not increase the risk of PM. In contrast, the interval time from biopsy to liver transplant among patients in the study by Heimbach et al. varied considerably with the median time from biopsy to surgery being over 100 days and with some intervals being longer than a year, which may explain the observed higher incidence of PM among those who had positive TP biopsies in that patient cohort.

To identify occult PM that might have been associated with TP biopsy but not grossly visible at the time of surgery and to account for a longer time interval from TP biopsy, we also assessed the cumulative incidence of PM at one and two years after surgery based on follow-up imaging and subsequent abdominal surgeries. In a competing risk analysis, no difference in the cumulative incidence of PM was found between patients who underwent TP biopsy and those who did not, demonstrating no association in the long term. Additionally, the biopsy result (positive vs. negative) had no association with the incidence of PM.

It is possible that the patients in our cohort have less advanced disease compared to the cohort of patients in the study by Heimbach et al., which included patients with localized, unresectable hilar cholangiocarcinoma and were enrolled in the Mayo Clinic transplant protocol. In our cohort, 37.4% ultimately had R0 resection and 21.7% had R1 resection. However, 41% ultimately were found to be unresectable at the index surgery, and the proportion of patients who had R0 resection, had R1 resection or were unresectable did not significantly differ among those who underwent TP biopsy vs. those who did not. This further suggests that one of the drivers of PM could be the underlying tumor biology and the presence of advance disease rather than TP biopsy.

Some studies have reported that peritoneal-violating interventions, such as PTBD, in PHC may increase the risk of peritoneal dissemination while larger studies have reported no significant association between PTBD and PM.^8–11 In our cohort, slightly more patients who underwent TP biopsy also had pre-operative PTBD compared to those that did not undergo TP biopsy. As such, it is possible that PTBD is a confounding factor in our study. Nevertheless, our results are consistent with studies of other TP interventions in cholangiocarcinoma that have failed to find an association with clinically significant disease recurrence. Prior studies have suggested that peritoneum violating interventions like PTBD increases the risk of seeding metastases and shortens OS compared to endoscopic drainage.^26,27 A study of 320 patients who underwent resection for PHC comparing 168 who underwent PTBD vs 152 who received endoscopic biliary drainage (EBD) before operation showed that percutaneous transhepatic biliary drainage increases the incidence of seeding metastasis and shortens the postoperative survival in patients with PHC.²⁷ However, another study at two large institutions comparing outcomes for patients with PHC who underwent preoperative EBD vs. preoperative PTBD found that OS and recurrence-free survival did not differ between groups. Furthermore, no patients who underwent percutaneous drainage developed catheter tract metastases as their primary recurrence, suggesting that while tumor seeding from TP intervention may occur, it is not the driver of disseminated disease.¹⁰

Limitations

This study is limited by its retrospective nature and small sample size. Despite this, the patient groups did not vary significantly in demographic, clinical, or therapeutic characteristics examined. The rarity of the disease and small number of events limits the feasibility of a larger randomized trial, especially since many potential subjects would only be identified after a tissue diagnosis had already been obtained. Furthermore, establishing a tissue diagnosis through biopsy in resectable patients prior to surgery is not always necessary. An additional limitation is the use of radiographic imaging to identify PM after the index surgery. The sensitivity of cross-sectional imaging for PM is roughly 84%^28–30 and therefore, some cases of PM may have been missed. In addition, it is also possible that advances in cross-sectional imaging could have made it more likely to detect PM among patients diagnosed in the latter years compared to those diagnosed in the early 1990s. Of the 243 patients that did not have peritoneal metastases (PM) at their index surgery, 72 eventually developed PM. Our data suggest that more cases of PM were detected in the latter years with majority of the PM cases identified between the years 2000 and 2015 (65/72, 90.3%), which is not surprising given the technological advances in imaging. However, the limitations of radiographic assessment of PM should apply equally to all patient cohorts regardless of biopsy type.

Conclusion

Peritoneal metastases occur in a substantial percentage of patients with PHC. In patients undergoing surgery for PHC, TP biopsy was not associated with a higher incidence of PM at surgery or cumulative incidence of PM at one or two years after surgery. Future investigations into novel clinical or genomic biomarkers of PHC may help predict which patients will develop peritoneal involvement. Our results suggest the risk of peritoneal seeding from TP biopsies may not be as high as case studies suggest, and patients who undergo TP biopsies should be considered candidates for the same treatment modalities as those who do not.