Biopsy of non-tumor sites after biopsy of a colorectal cancer is not associated with metachronous cancers: a case-control study

Abstract

Background and aims:

Recent research has demonstrated biologic plausibility for iatrogenic tumor seeding via colonoscopy as a cause of metachronous colorectal cancers (CRC). This study evaluated the association between biopsy of non-tumor sites after CRC biopsy and risk of metachronous CRC in a large community-based healthcare organization.

Methods:

This was a retrospective case-control study of adults with an initial CRC diagnosed by colonoscopy between January 2006 and June 2018 who underwent curative resection. Cases developed a second primary (metachronous) CRC diagnosed 6 months to 4 years after the initial CRC, and were matched by age, sex, diagnosis of inflammatory bowel disease, race, and ethnicity with up to 5 controls without a second CRC diagnosis. The exposure was biopsy in the colonic segment of the metachronous CRC (or corresponding segment in controls) after tumor biopsy, ascertained with blinding to case status. Associations were evaluated using conditional logistic regression and adjusted for potential cofounders.

Results:

Among 14,119 patients diagnosed with an initial CRC during colonoscopy, 107 received a second CRC diagnosis. After exclusions for recurrent or synchronous CRC, 45 cases and 212 controls were included. There was no significant association between biopsy of non-tumor sites after initial CRC biopsy and risk of metachronous CRC in the segment of the additional biopsy site (adjusted odds ratio 2.29, 95% CI 0.77–6.81).

Conclusions:

Metachronous cancers are not significantly associated with biopsy of non-tumor sites after biopsy of the primary cancer. Although the sample size does not allow definite exclusion of any association, these findings do not support iatrogenic tumor seeding as a common risk factor for metachronous CRC.

Graphical Abstract

INTRODUCTION

Despite effective screening and prevention strategies (1), colorectal cancer (CRC) remains the third leading cause of cancer-related death in the United States (2). To maximize the protective effect of screening colonoscopy, there has been increasing attention to colonoscopy quality with a goal of reducing post-colonoscopy colorectal cancers (PCCRC), i.e. a CRC that develops after a colonoscopy in which CRC is not detected (3). Known modifiable etiologies of PCCRC include missed lesions and incomplete polypectomy during the index colonoscopy, alongside non-modifiable factors such as rapidly growing lesions (4–7).

There is some overlap between risk factors for PCCRC and metachronous CRC. While defined as a second primary CRC detected after a colonoscopy in which CRC was found, metachronous CRC is similar to PCCRC in that the second cancer occurs in a site where cancer was not detected at the index colonoscopy. Recently, a proof-of-concept study from the Netherlands demonstrated that iatrogenic tumor seeding during a colonoscopy may also lead to metachronous CRC(8). In the study, metachronous CRC was defined as a second, primary CRC developing within 6 months to 3.5 years after resection of the initial CRC. Backes et al. demonstrated that viable tumor cells contaminate the endoscopic working channel and instruments following a tumor biopsy, which can be successfully cultured into an organoid. The authors hypothesized that these tumor cells can be implanted with post-tumor manipulation (biopsy) of non-malignant colorectal mucosa and confirmed identical molecular signatures of primary and metachronous tumors in 3 patients with suspected tumor seeding. These findings serve as the highest quality evidence to date for the possibility that tumor seeding may be a cause of metachronous CRC, with an estimated risk between 0.3–0.6% among patients who had an endoscopic intervention in a separate, non-malignant site following biopsy of a primary CRC. However, risk estimation in this study was limited by the small number of metachronous CRCs and the selected patient population (limited to 4 hospitals), creating ambiguity regarding whether this theoretical mechanism substantially contributes to the absolute and relative risks of metachronous cancers.

The present case-control study investigates the association between biopsy of non-tumor sites after biopsy of an initial CRC and the development of metachronous CRC in the biopsied colonic segment within a large, community-based integrated healthcare organization and a comprehensive cancer registry.

METHODS

Hypothesis

The study hypothesis posed that among patients diagnosed with CRC on index colonoscopy, biopsy of non-tumor sites after tumor biopsy would be associated with an increased risk of developing a second primary, or metachronous, CRC at or near the biopsied site.

Study design and population

This retrospective matched case-control study was performed using a cohort of adult Kaiser Permanente Northern California (KPNC) health plan members who were diagnosed with CRC followed by complete resection between January 1, 2006 and June 30, 2018. KPNC is an integrated health care delivery organization that serves approximately 4.5 million current members across 21 medical centers and 262 medical offices in urban, suburban, and semi-rural regions throughout Northern California (9). Membership is demographically diverse and similar in most characteristics to the region’s underlying census demographics (10).

The case-control design was selected over a prospective one due to its feasibility, given the very low estimated incidence of mechanical tumor seeding. For example, using a tumor seeding incidence of 0.6% in at-risk patients and 0% in control patients, a total sample size of 2,606 patients with CRC would be required for a prospective cohort study – challenging feasibility even within a large, integrated healthcare setting.

The study was approved by the KPNC Institutional Review Board and conducted within the National Cancer Institute-funded Population-based Research to Optimize the Screening Process (PROSPR 2) Consortium (UM1 CA222035). The listed authors had sole responsibility for study design, data collection, manuscript drafting, and submission for publication.

Study subject eligibility criteria

Individuals were eligible for the study if they were adult KPNC health plan members age ≥18 years, diagnosed with an initial CRC by colonoscopy with biopsy (referred to as the index colonoscopy) between January 1, 2006 and June 30, 2018, and managed with subsequent complete endoscopic or surgical resection. Patients were excluded from the study if they met any of following criteria: 1) had gaps in membership between the initial CRC diagnosis and metachronous CRC diagnosis (or reference date in controls) of >90 days in any calendar year; 2) had a hereditary CRC syndrome diagnosis based on International Classification of Disease (ICD-9 and ICD-10 codes) prior to the metachronous CRC diagnosis (or reference date in controls); or 3) the index examination extent did not reach the metachronous CRC site (or reference site in controls) raising the possibility of synchronous CRC (i.e. a second CRC was present at the same time as the first CRC diagnosis, but not detected because that area of the colon was not examined during the index colonoscopy).

Data sources and definitions

Demographic and clinical data were extracted from KPNC electronic health records and validated electronic cancer registry, demographic, and membership databases. Cancer diagnoses and characteristics were derived from the KPNC cancer registry, which reports to the Surveillance, Epidemiology, and End Results (SEER) program and maintains a >98% population-based completeness standard verified by random auditing (11), using ICD for Oncology, 3^rd Edition codes (Supplemental Table 1). Colonoscopy procedures were identified using Current Procedural Terminology codes, ICD-9 and ICD-10 procedure codes, Healthcare Common Procedure Coding System codes and KPNC-specific local codes (Supplemental Table 2). Index colonoscopy characteristics including examination extent, bowel preparation quality, tumor biopsy locations, non-tumor manipulation locations, polyp removal method and completeness, and pathology results of non-tumor biopsies and/or polypectomies were extracted by manual review of colonoscopy, pathology, surgical and imaging reports. Endoscopic or surgical resection of the index CRC was confirmed by chart review of colonoscopy and surgical operative reports.

Identification of cases and controls and outcome ascertainment

The primary outcome (case/control status) was development of a second, metachronous CRC using the cancer registry. Cases were defined as all KPNC members ages 18 or older who had a second CRC diagnosed within 6 months to 4 years of the initial CRC diagnosis date, and subsequent to surgical or endoscopic resection of the initial CRC. This definition was derived from PCCRC criteria and is similar to recent literature investigating tumor seeding as a potential cause for metachronous CRC (3,8). The metachronous CRC diagnosis and diagnosis date were confirmed by manual review of records to verify location and to exclude cases of suspected recurrence (second CRC at anastomotic site or at a site where a malignant polyp was previously resected endoscopically) and/or synchronous CRC (second CRC in location that was not reached at the index colonoscopy).

Controls were defined as all KPNC members age 18 or older who did not have a second CRC diagnosed within 4 years of the initial surgically or endoscopically resected CRC. Up to 5 controls were individually matched to each case using the following criteria: age (within 5 years of the matched case), sex, race, ethnicity, preceding inflammatory bowel disease (IBD) diagnosis, and follow-up interval from the initial CRC diagnosis within 1 year of metachronous CRC diagnosis of the matched case. The initial CRC site for controls was also restricted to colonic segments outside of the reference metachronous CRC site of matched cases. Controls were verified through manual record review to confirm tumor location, biopsy during the index colonoscopy, and subsequent surgical or endoscopic resection.

Exposure ascertainment

The exposure of interest was biopsy and/or polypectomy (also referred to as “biopsy”) after tumor biopsy during colonoscopy at the reference site. The reference site was defined as the colonic segment where metachronous CRC was subsequently diagnosed in cases and the corresponding colon segment in the matched controls. This exposure definition was guided by a theoretical randomized controlled trial design, where the hypothetical intervention would be biopsy or polypectomy at a separate, tumor-free colonic location (the “reference site”) following biopsy of the initial CRC and corresponding outcome of metachronous CRC development at the reference site. While post-tumor biopsies outside the reference site also carry the potential risk of tumor seeding, these would not match the case outcomes. Furthermore, as the mechanism suggested by prior literature was contamination of the endoscopic working channel and subsequently introduced instruments (8), the specific instrument used for biopsy or polypectomy was not factored into the exposure.

Among potential cases and matched controls, exposure status was ascertained by abstractors blinded to case/control status. All additional biopsies and polypectomies after tumor biopsy were recorded, noting biopsy/polypectomy locations, pathology results, and methods of polyp removal (if applicable). The sequence of biopsies was determined by a combination of colonoscopy and pathology reports, given the limitations of utilizing colonoscopy reports alone. For separate biopsy specimens submitted to pathology, the general protocol at KPNC is that each is labeled in the sequence of collection (e.g., bottle A, bottle B, bottle C). Only non-malignant specimens that followed biopsy of a colorectal adenocarcinoma were recorded as potential exposures. For instance, if bottle B contained adenocarcinoma, then only the biopsy at bottle C was considered a possible exposure. If there was more than 1 pathology result per colonic segment, the most advanced histology was recorded. If there was more than 1 polypectomy method per colonic segment, the most advanced method of resection was recorded (i.e., snare recorded over cold forceps polypectomy). Additional covariates extracted from index colonoscopy reports included examination extent (e.g., to cecum), completeness of any polyp resection, and quality of bowel preparation. To ensure inter-rater reliability, both extractors met weekly during the exposure ascertainment process to review overlapping assignments from the study cohort. If the manipulated colonic location was reported in centimeters from anal verge, conversion to a corresponding colonic segment was estimated using recommended anatomical measurements (12,13).

Statistical analysis

Baseline demographic and clinical characteristics of cases and controls are described. Distributions of categorical and ordinal variables between cases and controls were evaluated using chi-squared (χ²) distributions. The a priori analysis evaluated the association between biopsy of the reference site after biopsy of a colorectal adenocarcinoma and development of metachronous CRC in the reference site using conditional logistic regression with calculation of odds ratios and 95% confidence intervals. Odds ratios were used as an estimate of risk given the relative rarity of the outcome.

Additional sensitivity analyses were performed excluding patients with incomplete polyp removal at the reference site, accompanied by adjustment for bowel preparation quality and extent of examination. Reference site biopsy and polypectomy characteristics among cases and controls were described using descriptive statistics and proportion comparisons were evaluated using χ² tests.

RESULTS

Characteristics of the study cohort

During the study period, a total of 945,532 colonoscopies were performed among adult KPNC members and 14,119 CRC diagnoses were made by colonoscopy with biopsy (Figure 1). Among these patients, 107 patients were diagnosed with a second CRC within 6 months to 4 years of the initial diagnosis. One patient was excluded due to lack of pathologic confirmation, 54 were excluded as likely recurrences during manual chart review (second CRC at the same site as the initial CRC), two patients were excluded as likely synchronous CRC (likely present at time of first cancer, given location in colonic segment not adequately evaluated at index examination), and 3 were excluded as the initial CRC was unresectable due to metastatic disease at diagnosis. Among the 47 remaining eligible cases, 45 were successfully matched to 212 controls in the final study cohort (2 cases were unable to be matched to controls within criteria).

Figure 1.

Cohort development diagram

The median age for cases and controls was 71.8 and 71.6 years, respectively; the majority were male (61.9%) and non-Hispanic White (76.7%)(Table 1). In the cohort, 37.8% of cases and 34.9% of controls carried a prior diagnosis of IBD (p=0.714). The proportions of patients with a family history of CRC were 24.4% in cases and 20.8% in controls (p=0.584). On index colonoscopy, 78.6% of examinations reached the cecum, and only 4.7% reported poor bowel preparation (of note these were all examinations with cancers thus some were obstructing cancers, precluding complete examinations).

Table 1.

Baseline characteristics of cases and controls

	Total	Cases	Controls	P-Value
	N (%)	N (%)	N (%)
Total	257	45	212
COHORT CHARACTERISTICS
Age (years)
Median (Interquartile Range)	71.6 (63.4–77.9)	71.8 (63.2–77.7)	71.6 (63.5–78.1)	0.943
Gender				0.957
Female	98 (38.1)	17 (37.8)	81 (38.2)
Male	159 (61.9)	28 (62.2)	131 (61.8)
Race/Ethnicity				0.915
Hispanic	33 (12.8)	6 (13.3)	27 (12.7)
Non-Hispanic
Asian/Pacific Islander	14 (5.4)	3 (6.7)	11 (5.2)
Black	13 (5.1)	3 (6.7)	10 (4.7)
White	197 (76.7)	33 (73.3)	164 (77.4)
History of Inflammatory Bowel Disease	91 (35.4)	17 (37.8)	74 (34.9)	0.714
Family History of Colorectal Cancer	55 (21.4)	11 (24.4)	44 (20.8)	0.584
ENDOSCOPIC CHARACTERISTICS
Indication for Exam				0.388
Diagnostic	181 (70.4)	35 (77.8)	146 (68.9)
FIT-positive	51 (19.8)	5 (11.1)	46 (21.7)
Surveillance	12 (4.7)	3 (6.7)	9 (4.2)
Screening	13 (5.1)	2 (4.4)	11 (5.2)
Extent of Exam				0.409
Cecum	202 (78.6)	33 (73.3)	169 (79.7)
Ascending colon	11 (4.3)	3 (6.7)	8 (3.8)
Hepatic flexure of colon	10 (3.9)	1 (2.2)	9 (4.2)
Transverse colon	6 (2.3)	1 (2.2)	5 (2.4)
Splenic flexure of colon	1 (0.4)	0 (0.0)	1 (0.5)
Descending colon	9 (3.5)	4 (8.9)	5 (2.4)
Sigmoid colon	11 (4.3)	3 (6.7)	8 (3.8)
Rectum, NOS	2 (0.8)	0 (0.0)	2 (0.9)
Unknown	5 (1.9)	0 (0.0)	5 (2.4)
Quality of Bowel Prep				0.983
Excellent	18 (7.0)	3 (6.7)	15 (7.1)
Good	162 (63.0)	27 (60.0)	135 (63.7)
Fair	36 (14.0)	7 (15.6)	29 (13.7)
Poor	12 (4.7)	2 (4.4)	10 (4.7)
Unknown	29 (11.3)	6 (13.3)	23 (10.8)
Non-Tumor Biopsy	81 (31.5)	14 (31.1)	67 (31.6)	0.948
Number of Non-Tumor Biopsies				0.231
0	176 (68.5)	31 (68.9)	145 (68.4)
1	48 (18.7)	9 (20.0)	39 (18.4)
2	18 (7.0)	2 (4.4)	16 (7.5)
3	8 (3.1)	0 (0.0)	8 (3.8)
4	1 (0.4)	0 (0.0)	1 (0.5)
≥5	6 (2.3)	3 (6.7)	3 (1.4)
COLORECTAL CANCER CHARACTERISTICS
Number of Primary Tumors				0.225
1	255 (99.2)	44 (97.8)	211 (99.5)
2	2 (0.8)	1 (2.2)	1 (0.5)
Adjuvant Radiation Therapy	13 (5.1)	2 (4.4)	11 (5.2)	1.000
Adjuvant Chemotherapy	67 (26.1)	11 (24.4)	56 (26.4)	0.784
Other Adjuvant Therapy	7 (2.7)	2 (4.4)	5 (2.4)	0.354
Interval Between Initial and Metachronous Cancer Diagnoses (years) *
Mean (Standard Deviation)	1.9 (1.0)	1.9 (0.9)	1.9 (1.0)	0.904
Median (Interquartile Range)	1.8 (1.1–2.8)	1.7 (1.1–2.7)	1.8 (1.1–2.8)	0.896
Range	0.1–3.9	0.5–3.7	0.1–3.9

^*For controls, this is the reference date of metachronous cancer for the matched case

One case and one control had synchronous primary tumors, thus there were 46 total tumors in the case group and 213 total tumors in the control group (Table 2). Tumor grades for the initial CRC diagnoses were more advanced in cases than controls (93.4% vs. 75.2% were undifferentiated or moderately to poorly differentiated, respectively; p=0.002). Histologically, 19.6% and 34.3% of initial CRC diagnoses in cases and controls, respectively, arose within polyps (adenocarcinoma in adenomatous polyp, villous adenoma, or tubulovillous adenoma). There were no statistically significant differences in tumor location or TNM staging for the initial CRC diagnoses between cases and controls. The mean time interval between initial and metachronous cancer diagnoses in cases was 1.9 years.

Table 2.

Characteristics of the initial colorectal cancer diagnoses for cases and controls

	Total	Cases	Controls	P-Value
	N (%)	N (%)	N (%)
Total Primary Tumors	259	46	213
Primary Tumor Location				0.226
Cecum	51 (19.7)	12 (26.1)	39 (18.3)
Ascending colon	42 (16.2)	3 (6.5)	39 (18.3)
Hepatic flexure of colon	13 (5.0)	3 (6.5)	10 (4.7)
Transverse colon	24 (9.3)	3 (6.5)	21 (9.9)
Splenic flexure of colon	9 (3.5)	1 (2.2)	8 (3.8)
Descending colon	13 (5.0)	5 (10.9)	8 (3.8)
Sigmoid colon	56 (21.6)	12 (26.1)	44 (20.7)
Rectosigmoid junction	12 (4.6)	1 (2.2)	11 (5.2)
Rectum, NOS	39 (15.1)	6 (13.0)	33 (15.5)
Primary Tumor Histology				0.070
Carcinoma & Adenocarcinoma, NOS	155 (59.8)	31 (67.4)	124 (58.2)
Adenocarcinoma in adenomatous polyp	82 (31.7)	9 (19.6)	73 (34.3)
Villous adenocarcinoma	1 (0.4)	0 (0.0)	1 (0.5)
Mucinous or mucin-producing adenocarcinoma	20 (7.7)	5 (10.9)	15 (7.0)
Signet ring cell carcinoma	1 (0.4)	1 (2.1)	0 (0.0)
Primary Tumor Grade				0.002
Well differentiated	24 (9.3)	0 (0.0)	24 (11.3)
Moderately differentiated	175 (67.6)	34 (73.9)	141 (66.2)
Poorly differentiated	24 (9.3)	6 (13.0)	18 (8.5)
Undifferentiated	4 (1.5)	3 (6.5)	1 (0.5)
Unknown	32 (12.4)	3 (6.5)	29 (13.6)
Primary Tumor TNM Stage				0.941
0	34 (13.1)	5 (10.9)	29 (13.6)
1	77 (29.7)	13 (28.3)	64 (30.1)
2	76 (29.3)	15 (32.6)	61 (28.6)
3	60 (23.2)	11 (23.9)	49 (23.0)
4	9 (3.5)	2 (4.3)	7 (3.3)
Unknown	3 (1.2)	0 (0.0)	3 (1.4)
Resected Surgically	249 (96.1)	42 (91.3)	207 (97.2)	0.081

Post-tumor biopsy/polypectomy characteristics at the reference site

Narrative details are provided in Table 3 for the 5 cases with the exposure of biopsy at the reference site after CRC biopsy at index colonoscopy, as these are most likely to represent cases of potential mechanical tumor seeding. Two out of the 5 cases (case 4 and 5) had a concomitant diagnosis of IBD. Thus, both may have had the potential for a field effect of increased cancer risk in other colon locations, though one had quiescent disease in remission (case 4) and the other had mild proctitis limited to the rectum that was diagnosed at the same time as the metachronous CRC (case 5). Case 4 had a normal pathology result at the reference site, as this was a random biopsy taken for surveillance of quiescent Crohn’s colitis, performed after biopsy of an ulcerated cecal lesion (the initial CRC). Polyp sizes in these 5 cases ranged up to 10 mm and were identified as either tubular adenomas or hyperplastic polyps on pathology.

Table 3.

Additional Details of Cases with Potential Tumor Seeding^*

Case	Primary Tumor Location	Metachronous Tumor Location	Interval Between Diagnoses	Reference Site Findings at Index Colonoscopy
1	Cecum	Descending colon	2.57 years	Five 5mm tubular adenomas, snare excised, complete
2	Sigmoid colon	Hepatic flexure	2.97 years	One 6mm hyperplastic polyp, snare excised, complete
3	Cecum	Descending colon	1.69 years	Two 5mm tubular adenomas, snare excised, complete
4	Cecum	Transverse colon	2.99 years	Normal (biopsy done for history of Crohn’s colitis, which was in remission)
5	Ascending colon	Transverse colon	0.61 years	Three 4–10mm polyps (tubular adenomas and hyperplastic polyps), snare excised, complete. Regarding IBD history, mild proctitis was subsequently diagnosed at follow-up colonoscopy (when metachronous CRC was also diagnosed).

^*Development of metachronous colorectal cancer at reference site after post-tumor biopsy at index colonoscopy

Additional details relating to biopsy and polypectomy characteristics are provided in Supplementary Table 3. There were no statistically significant differences between cases and controls in terms of the extent of polyp removal, method of removal, and polyp histology.

Exposure analysis

Within the case group, 5 out of 45 patients (11.1%) had an exposure of post-tumor biopsy and/or polypectomy at the reference site (segment where metachronous CRC developed for cases or matching segment for controls), compared to 13 out of 212 controls (6.1%). However, patients with biopsy of non-tumor sites after tumor biopsy did not have a significantly higher risk of metachronous cancer diagnosed within the biopsied colonic segment compared with controls (unadjusted odds ratio: 2.03; 95% confidence interval: 0.70–5.91, Table 4). Adjusting for bowel preparation adequacy, extent of examination, and completeness of polyp removal did not substantially change the point estimate or statistical significance (adjusted odds ratio: 2.29; 95% confidence interval 0.77–6.81).

Table 4.

Odds of having endoscopic manipulation after colorectal adenocarcinoma biopsy at a separate cancer-free reference site^* in cases and controls

	Cases	Controls	Odds Ratio [95% Confidence Interval]
	N	N	Unadjusted	Adjusted **
Total	45	212
Exposure status
No biopsy at reference site after tumor biopsy	40	199	1.00 [reference]	1.00 [reference]
Biopsy performed at reference site after tumor biopsy	5	13	2.03 [0.70, 5.91]	2.29 [0.77, 6.81]

^*Site of metachronous cancer in cases, and the corresponding site in matched controls

^**Adjusted for completeness of exam, quality of bowel preparation and completeness of polyp removal

DISCUSSION

This study in a large integrated healthcare setting found that metachronous CRC diagnoses after primary cancer diagnosis were rare, and that biopsy of non-tumor sites after biopsy of an initial CRC (during the same colonoscopy) was not significantly associated with metachronous CRC in the biopsied colonic segment. These findings do not support the concept that biopsy of non-malignant sites after biopsy of a CRC confers a meaningfully higher risk of metachronous cancer, despite biologic plausibility (8).

These results add to the scarce literature on a possible iatrogenic etiology of metachronous CRC. While the theoretical risk of mechanical tumor seeding appears low based on prior literature primarily comprised of mammalian models and case reports (14), the recent proof-of-concept study established a viable biologic mechanism via proven contamination of the colonoscope’s working channel and associated instruments, with tumor cells demonstrating proliferative potential (8). Backes et al. further verified likely causality through genomic sequencing, which showed identical molecular signatures between the initial and metachronous CRC that differed from the separate polyp removed at index colonoscopy. Clinical studies evaluating the risk of mechanical tumor seeding are challenging, due to the rarity and low absolute risk of metachronous cancer. Prior estimates of metachronous CRC risk in the literature ranged from 1.1% at 3 years to 3.1% at 10 years in a Dutch cancer registry study, though this was limited by lack of supplemental medical record/colonoscopy data to adequately exclude synchronous cancers or anastomotic recurrences and may have been influenced by exam quality (15). In a more recent systematic review and meta-analysis, the cumulative incidence of non-anastomotic metachronous CRCs was 2.2% over follow-up periods exceeding 16 years (16). However, CRCs developing after 4 years are likely to be new cancers unrelated to modifiable factors at the index colonoscopy (3). A small study of patients who had synchronous polypectomy undergoing surgical resection of an initial CRC in Singapore suggested comparable rates of metachronous CRC in those who had the synchronous polypectomy site resected compared to those who had polypectomy sites left in-situ after 5 years of follow-up (1.1% and 1.5%, p=0.795) (17). However, the study was limited in scale with only 6 overall cases of metachronous CRC. Utilizing data from a large, community-based integrated healthcare organization, the current study identified 45 matched cases of metachronous CRC to estimate the potential association between biopsy of a cancer followed by biopsy of non-tumor sites and development of a metachronous CRC in the same colonic segment as that biopsy. Despite the recently demonstrated biological plausibility of iatrogenic tumor seeding, our findings suggest that it is unlikely that endoscopic practices have been significantly contributing to the development of metachronous CRCs.

The study design has several strengths and limitations. A notable strength is the large and representative patient population with longitudinal electronic databases spanning more than 12 years across 21 medical centers. The case-control design efficiently studies rare outcomes such as metachronous CRC. Although colonoscopy can be inaccurate for localization of tumors (erroneous in approximately 20% of cases) (18), we manually reviewed operative and imaging reports for each patient to optimize report locations of both initial and metachronous tumors. This also addressed potential inaccuracies from relying on cancer registry classifications alone for tumor localization. The process of manual review also facilitated exclusion of secondary CRC that likely represented cases of recurrent or synchronous CRC. We mitigated potential abstractor bias regarding exposure status by using abstractors blinded to case-control status. To reduce the risk of confounding, we also matched on several criteria associated with cancer risk, including sex, race, ethnicity, age, IBD diagnosis, and follow-up time interval (to provide comparable time intervals for cancer development between initial and metachronous CRC diagnoses in cases and controls). There were no patients with hereditary CRC syndromes in the study. However, accuracy may be limited due to dependence on ICD code diagnoses for IBD and hereditary syndromes. There were several limitations that preclude definitive statements regarding causation in the 5 cases where exposure criteria were met and tumor seeding considered a possibility. First, limitations of biopsy site localization limited the ability to precisely state that the biopsy and metachronous cancer occurred at the exact same site; to address this as fully as possible we used colon segments, blinded ascertainment, and standardized data abstraction protocols. Second, it is possible that metachronous tumors occurred due to neoplastic progression of incompletely resected polyps, missed lesions on the index colonoscopy, or an etiologic field effect (3,19). Thus, people who have more polyps are both at higher likelihood of getting biopsies and higher risk of future cancers at or near those sites. However, adjustments for bowel preparation adequacy, examination extent, and completeness of polyp removal did not show substantial changes in risk estimates (3,5–7). Third, we did not have information from genetic analyses; this would require molecular analysis of the initial tumor, metachronous tumor, and removed polyp, which was outside the scope and resources of the current study. Fourth, the study primarily relied on sequence of pathology specimens to determine whether additional biopsies took place after colorectal adenocarcinoma biopsy. This may have missed a subset of biopsies that were taken after adenocarcinoma biopsy but placed in a preceding specimen bottle, or the possibility that samples taken from separate locations were collected in the same specimen bottle. Fifth, the study design, with blinded ascertainment status, biased towards a higher odds ratio in cases compared with controls, e.g. if the extent of the index colonoscopy in controls did not reach the reference site leading to an automatic “negative” exposure status (this was uncommon). However, as there was still no significant association found, this bias did not influence the overall conclusion. Finally, the small sample size, which is due to the relative rarity of metachronous CRC, does not completely exclude the possibility of a type II error. We note that the OR point estimate was in the direction of an association, but did not reach statistical significance. However, future studies will likely also be hampered by power considerations, given the rarity of the outcome.

While informative, the results of this study do not completely exclude the possibility that tumor seeding can still occur in rare instances. Practices to mitigate this risk, such as deferring manipulation of a suspected CRC until the end of a procedure, flushing the channel after biopsy and before other manipulations, and changing biopsy forceps after biopsy of a cancer are still reasonable, though without strong evidence that they are strictly necessary.

In conclusion, metachronous cancers are rare after primary CRCs and are not associated with biopsy of non-tumor sites after biopsy of the primary cancer. Although sample size cannot exclude any association, these clinical findings do not support iatrogenic tumor seeding as a common risk factor for metachronous CRC.

What You Need to Know:

BACKGROUND

Iatrogenic tumor seeding during colonoscopy may cause metachronous colorectal cancer. We evaluated the association between biopsy of non-tumor sites after colorectal adenocarcinoma biopsy and the development of metachronous cancer.

FINDINGS

There was no statistically significant association between biopsy of non-tumor sites after initial colorectal cancer biopsy and development of subsequent metachronous colorectal cancer in the region of the biopsied site.

IMPLICATIONS FOR PATIENT CARE

Despite the biological plausibility of iatrogenic tumor seeding, these findings suggest it is unlikely that endoscopic practices have been significantly contributing to the development of metachronous colorectal cancers.

Abbreviations:

ADR

adenoma detection rate

CRC

colorectal cancer

IBD

inflammatory bowel disease

ICD

International Classification of Disease

KPNC

Kaiser Permanente Northern California

PCCRC

post-colonoscopy colorectal cancer

PROSPR 2

Population-based Research to Optimize the Screening Process

SEER

Surveillance, Epidemiology, and End Results