Context.—

Tumor contaminants were incidentally noted in frozen section margins of oropharyngeal squamous cell carcinoma.

Objective.—

To estimate the frequency of tumor contaminants in frozen section slides of patients who underwent surgery for pharyngeal cancer, and to characterize the surgical and pathologic context of these incidents.

Design.—

A retrospective search was conducted to identify pharyngeal resections from 2016 to 2022. Surgical pathology, operative reports, and frozen section slides were reviewed. Preanalytical phase tumor contaminants were defined as tumor contaminants that occurred in frozen section slides with or without occurrence in permanent slides.

Results.—

Eighty-one pharyngeal resections with intraoperative tumor bed margins for squamous cell carcinoma were identified. These included 308 tumor bed margins represented in 641 slides. Preanalytical contaminants occurred among 9 patients (11.1% of all and 21.4% of robotic surgeries) and in 3.8% of the 308 intraoperative tumor bed margins. A statistically significant association was found between contaminants and larger tumor size (Student t test, P = .04) and surgical approach (robotic versus open oropharyngectomy: Fisher exact test, P < .001). All patients with contaminants had intraoperative tumor disruption. Two frozen section deferrals (0.6%) and 2 discrepancies with final diagnosis (0.6%) attributed to contaminants were identified; however, clinical or surgical management was not affected in any patient.

Conclusions.—

Preanalytical contaminants may cause confusion in intraoperative margin assessment. They are more likely to occur in margins of nonkeratinizing squamous cell carcinoma resected by transoral robotic surgery if there is intraoperative tumor disruption. Rarely, preanalytical contaminants lead to frozen section deferral or discrepancy with final diagnosis.

Intraoperative frozen section (FS) is routinely used for margin assessment in head and neck cancer surgery. The accuracy of determining positive margins with FS is found to be 97% to 98%.1–3  FS aids in complete tumor extirpation and thus, reducing recurrence, improving survival, and reducing further adjuvant management.1 

Robotic surgery was first used in the field of neurosurgery, after which it has been incorporated to other surgical fields like abdominal, gynecologic, and urologic, among others. Transoral robotic surgery (TORS) is a minimally invasive procedure that was developed in the 2000s for head and neck cancers.4  TORS aids surgeons in getting better access to tumors, greater visibility, and reduced blood loss during surgery with the help of a robot-assisted device.5  Although TORS has many advantages, there are a few known disadvantages, namely lack of sensory tactile feedback and costly setup.6 

There are several well-known pitfalls in the assessment of intraoperative tumor bed margins in the head and neck region.7–10  To our knowledge, there are very few publications on the accuracy and precision of intraoperative FS for margin assessment focused specifically in TORS for oropharyngeal squamous cell carcinoma (OPSCC).11 

We incidentally found detached tumor fragments away from the primary tumor in FS glass slides of tumor bed margins in patients undergoing TORS. In this study, we aim to estimate the frequency of tumor contaminants in FS slides of patients who underwent surgery for pharyngeal cancer, and to characterize the surgical and pathologic context of these incidents occurring in patients undergoing open surgery and TORS for OPSCC.

Following institutional review board approval, a retrospective study of patients undergoing surgery for oropharyngeal tumors from 2016 to 2022 was conducted at a single institution. Patients were included if they had TORS or open resection for an OPSCC (either primary or recurrent). Eligible patient charts were reviewed, and data were collected on demographics including age, sex, race, smoking status, alcohol consumption, and body mass index (BMI). Additional data were collected on tumor histologic type, p16 status, and surgical factors including tumor size, tumor node metastases (TNM) classification, perineural invasion (PNI), and lymphovascular invasion (LVI). Operative notes were reviewed to determine if the tumor was resected en bloc versus piecemeal, was transected intraoperatively, or had close/positive margins. Finally, data on perioperative complications were collected, including intraoperative bleeding, unanticipated return to the operating room, and neck infection.

FS slides were screened by a head and neck pathologist (HNP; MALG). Problematic tumor fragments were defined as fragments of tumor with confounding potential (possible contaminants). Preanalytical contaminants (PACs) were defined as detached tumor fragments in FS slides of tumor bed margins away from the primary tumor, with or without occurrence in permanent slides. After FS slides with contaminants were identified they were reviewed by 2 additional HNPs (DL, RC)—along with permanent slides and primary tumor from the main resection specimen—blinded to surgery type. A slide was considered indeterminate if the HNPs disagreed with each other on whether the tumor fragment represented a contaminant. Final pathology reports, permanent slides, and intraoperative reports were retrospectively reviewed and discrepancies were annotated.

Statistical analysis was performed with Prism v10.3.0 (GraphPad Software, Boston, Massachusetts). Continuous variables were assessed with mean and standard deviation, and categorical variables were assessed with frequency and percentage. Age, BMI, tumor size, and cautery levels were compared between patients with and without preanalytical contaminants. Each variable was tested for normality with the Shapiro-Wilk test and compared by using the Welch t test (if Shapiro-Wilk P > .05) or Mann-Whitney U test (if Shapiro-Wilk P < .05). Sex, race, smoking status, alcohol consumption, p16 status, diagnosis, tumor stage, PNI, LVI, perioperative complications, adjuvant treatment, cautery levels, and histology of tumor were compared between patients with and without preanalytical contaminants by using Fisher exact test. Statistical significance was defined as P < .05.

One hundred twenty-seven patients underwent surgery for OPSCC from 2016 to 2022. Patients whose surgery included FSs (n = 81) were included in the study, of whom 42 underwent TORS and 39 underwent open oropharyngectomy. Of 308 tumor bed margins, 289 were submitted for FS, of which 128 (44.3%) originated from TORS and 161 (55.7%) originated from open oropharyngectomies. PACs occurred in 11 of 289 intraoperative tumor bed margins (3.8%) from 9 of the 81 patients (11.1%). These represent 9 of the 42 TORS surgeries (21.4%).

The 2 surgical cohorts (patients undergoing TORS versus open pharyngectomy) were demographically similar except for a statistically significant difference in p16 status (Table 1). TORS patients were more likely to have p16-positive tumors (P =. 007). There were no significant differences between smoking, alcohol consumption status, and BMI.

Table 1.

Study Population Demographics

Study Population Demographics
Study Population Demographics

Abbreviation: TORS, transoral robotic surgery.

The indications for TORS versus open pharyngectomy were different. Among the patients undergoing TORS, 12 (28.6%) had carcinoma of unknown primary and 30 (71.4%) had tumors identifiable on examination or imaging. Fifty percent of the tumors in TORS cases had pathologic tumor classification of T2. In comparison, most patients undergoing open oropharyngectomies had salvage surgeries (n = 27, 69.2%) and T4-stage tumors (n = 17, 43.6%) with 46.1% (n = 18) of the tumors having a size of more than 4 cm. Most of the patients in the TORS cohort had negative PNI (n = 31, 73.8%) and negative LVI (n = 23, 54.8%). PNI was identified in 51.3% (n = 20) and LVI was identified in 41.2% (n = 16) of the open surgery cohort.

Intraoperative complications occurred in 4.7% (n = 2) of TORS cases and included bleeding and chyle leak. Intraoperative complications (bleeding) occurred in 2.6% (n = 1) of the open cases. The cautery level was set higher than 30 in 26.2% (n = 11) of the TORS cases versus 7.7% (n = 3) of the open oropharyngectomy cases. Use of higher cautery setting was significantly more frequent in the TORS cohort than in the open surgery cohort. The surgical factors for the patients undergoing TORS and open oropharyngectomy are detailed in Table 2.

Table 2.

Surgical and Pathologic Characteristics of the Study Groups

Surgical and Pathologic Characteristics of the Study Groups
Surgical and Pathologic Characteristics of the Study Groups

A discrepancy between frozen and final pathology was determined in 3 of 128 TORS FS margins (2.3%) from 2 patients. A discrepancy between the frozen and final pathology was observed in 3 of the 161 open pharyngectomy FS margins (1.8%). Of these, 2 discrepancies were false positives that occurred owing to the presence of PACs, both in the TORS cohort. These discrepancies were diagnosed as “positive for carcinoma” and “atypical” on FS. Two FS deferrals occurred that were attributed to PACs.

The FS slides of the study patients included 308 tumor bed margins, which were represented in 641 slides. Seventeen problematic tumor fragments (possible tumor contaminants) were noticed in FS slides, which included 15 from the TORS cohort and 2 from the open surgery cohort. Upon review by 3 HNPs, 6 of these problematic tumor fragments (4 in the TORS cohort and 2 in open oropharyngectomy cohort) were not deemed as a preanalytical contaminant, as a grossing (analytical phase) contaminant could not be excluded. Thus, 11 of the 17 problematic fragments were classified as true PACs by HNP consensus. The Figure, A through I, shows 3 examples of PACs in intraoperative margin slides along with their corresponding primary tumor.

Microscopic findings in 3 examples of tumor bed margins with tumor contaminants in tumor bed margin slides (A, B, D, E, G, and H) and corresponding primary tumors in the right column (C and F). Immunohistochemistry with p16 highlights the detached tumor fragment in a human papillomavirus–mediated carcinoma (I) (hematoxylin-eosin, original magnifications ×20 [A, D, and G], ×100 [C and F], and ×200 [B, E, and H]; original magnification ×200 [I]).

Microscopic findings in 3 examples of tumor bed margins with tumor contaminants in tumor bed margin slides (A, B, D, E, G, and H) and corresponding primary tumors in the right column (C and F). Immunohistochemistry with p16 highlights the detached tumor fragment in a human papillomavirus–mediated carcinoma (I) (hematoxylin-eosin, original magnifications ×20 [A, D, and G], ×100 [C and F], and ×200 [B, E, and H]; original magnification ×200 [I]).

Close modal

Upon retrospective review of the cases with preanalytical phase tumor contaminants, all had at least one of the following operative circumstances in common: (1) surgeon had to cut through primary tumor for removal or access (n = 5); (2) primary tumor had to be removed in piecemeal or with a positive deep margin (n = 7); (3) tumor biopsy samples were sent for FS prior to resection for primary tumor identification (n = 1); or (4) a tear occurred in the specimen at the tumor site during removal (n = 1). Characteristics of patients with preanalytical contaminants are detailed in Table 3.

Table 3.

Characteristics of Incidents Classified as Preanalytical Tumor Contaminants

Characteristics of Incidents Classified as Preanalytical Tumor Contaminants
Characteristics of Incidents Classified as Preanalytical Tumor Contaminants

Associations between occurrence of preanalytical contaminants with various demographic, surgical, and pathologic factors were assessed (Table 4). No demographic factors were found to be associated with occurrence of contaminants between the 2 cohorts (cohort with contaminants and cohort without contaminants), including age, sex, race, smoking, alcohol consumption, and BMI.

Table 4.

Cases With Tumor Contaminants Versus Cases Without Tumor Contaminants

Cases With Tumor Contaminants Versus Cases Without Tumor Contaminants
Cases With Tumor Contaminants Versus Cases Without Tumor Contaminants

Among the surgical factors, a statistically significant association was found between contaminants and tumor size (Welch t test, P = .04). The surgical approach used (TORS versus open oropharyngectomy) was associated with preanalytical contaminants (Fisher exact test, P = .002). No other associations were found between contaminants and other surgical factors including diagnosis (primary or recurrence), tumor stage, PNI, LVI, intraoperative cautery levels, perioperative complications, and adjuvant treatment. A statistically significant association was found between the histologic type of the tumor, that is, the nonkeratinizing type of SCC, which is often human papillomavirus (HPV) associated, and occurrence of contaminants (Fisher exact test, P < .001). There was no significant association between occurrence of contaminants and p16 status, a surrogate marker for HPV (Fisher exact test, P = .07).

FS has been used as an intraoperative diagnostic modality from the late 1890s.8  Achieving negative margins in HNC surgery is important as it affects survival and guides decisions relating to adjuvant treatment. With the advent of robotic surgery, there have been reports about tissue artifacts and tumor contaminants affecting specimen analysis in these surgeries.11,12  To our knowledge, this is the only study describing the occurrence of PACs in tumor bed margins in patients with OPSCC undergoing surgery as standard of care. In this multivariate study, we characterized the clinical and pathologic context of these occurrences.

Tissue contaminants and artifacts are not uncommon in surgical pathology practice despite standard protocols. Pathologists are trained to recognize and troubleshoot most types of contaminants and they do so in everyday practice. Most contaminants encountered in practice represent analytical phase cross-contaminants in permanent sections originating in the laboratory as gross instrument contaminants, cross-contaminants from a tissue processor, and histology water bath “floater” cross-contaminants. The rate of occurrence of these is very low (0.6%).13,14 

Preanalytical phase contamination occurs even less frequently, given the minimal risk of cross-contamination with current specimen collection standards. In some instances, however, extraneous tissue from the same patient may be present in a specimen container. We studied PACs occurring in FS slides of tumor bed margins for OPSCC, in which there is potential to interfere with intraoperative margin interpretation. We found PACs in 3.8% of all tumor bed margins for OPSCC but all PACs occurred in 9 patients who underwent TORS, corresponding to one-fifth of the TORS surgeries we evaluated. We inferred that our contaminants originated in the preanalytical phase because of their presence in FS slides of specimens received directly from the operating room and processed in the FS room, before the steps of routine processing (tissue processor, water bath, grossing table, and histology) during which analytical phase cross-contaminants typically occur.

For the 2 cohorts of patients undergoing TORS and open surgery for OPSCC, the age, sex, race, smoking, and alcohol status, and BMI were similar in both the groups. A statistically significant difference was found between the modality of treatment between the 2 groups.

While the demographics and baseline characteristics were comparable between the cohorts undergoing TORS and open surgery for OPSCC, patients undergoing TORS had a higher incidence of tumor contaminants than the open surgery group. Delving deeper, we explore potential factors contributing to this.

Moreover, all contaminants except 1 originated from TORS for nonkeratinizing, HPV-positive tumors. We contemplate whether contaminants occurred more commonly in this type of tumor because these tumors recapitulate the tonsillar crypt epithelium in which expression of intercellular tight junctions is different from that of surface keratinizing mucosa (less cohesive).15 

Upon retrospective review, an association was found between larger tumor size and occurrence of PACs, and all PACs occurred in cases where the tumor was disrupted during surgery in the form of piecemeal dissection or obtaining access by cutting through the tumor or complete excision or tear occurring at the tumor site or tumor biopsy specimens being taken intraoperatively. After all, preanalytical contaminants in the tumor bed margins of TORS may just be a consequence of the challenges related to patterns of spread of oropharyngeal tumors and the limited maneuverability in the anatomic region that makes TORS the preferred technique in these cases. We also pondered if contamination could be more likely in high cautery settings, as some cautery technologies have a been associated with significant tissue distortion on glass slides.16  Our analysis did not find association between cautery settings and occurrence of PACs. The TORS cohort had a higher number of patients with p16-positive status. However, on further analysis, no association was found between p16 positivity and occurrence of tumor contaminants.

While the occurrence of PACs may seem trivial, we believe their relevance should not be underestimated. In our cohort, we identified 1 false-positive margin and 1 equivocal (atypical) margin explained by pathologist interpretive error caused by PACs, which did not cause injury to the patients, but had the potential to prolong anesthesia duration and/or result in additional tissue being excised unnecessarily.

While tissue contaminants rarely cause human interpretive error,17  tissue contaminants may interfere with the performance of emerging technologies such as machine learning models. As these technologies emerge further study of the impact of PACs may be necessary. For example, a recent study demonstrated that the addition of contaminants induced false negatives and false positives, and under specific circumstances, resulted in a 97% false-positive rate in a modern machine learning model.18  Keeping in mind that contaminants have the potential to interfere with the development of these emerging technologies, this study is a first attempt to understand the occurrence of contaminants in this setting.

Finally, when considering the proportion of patients in which PACs occurred (21.4% of the patients who underwent TORS), these incidents may not be so uncommon. Regarding the correct approach for the FS pathologist reporting margins that may have PACs, we surmise that using descriptive terminology and stating “possible contaminant” is a reasonable option that gives the surgeon the opportunity to correlate with their findings in the tumor bed.

PACs occurred in a small subset of patients who underwent surgery for OPSCC. Most occur in the setting of TORS for a large HPV-mediated OPSCC with intraoperative disruption of tumors with nonkeratinizing histology. Rarely, PACs lead to FS deferral or discrepancy with final diagnosis. It is important for surgical pathologists and head and neck surgeons to recognize the possibility of PACs when assessing intraoperative margins to prevent possible confusion and further unnecessary treatment. These findings underscore the importance of communication between pathologists and surgeons.

1.
DiNardo
LJ
,
Lin
J
,
Karageorge
LS
,
Powers
CN.
Accuracy, utility, and cost of frozen section margins in head and neck cancer surgery
.
Laryngoscope
.
2000
;
110
(
10 pt 1
):
1773
1776
.
2.
Zarbo
RJ
,
Hoffman
GG
,
Howanitz
PJ.
Interinstitutional comparison of frozen-section consultation: a College of American Pathologists Q-Probe study of 79,647 consultations in 297 North American institutions
.
Arch Pathol Lab Med
.
1991
;
115
(
12
):
1187
1194
.
3.
Layfield
EM
,
Schmidt
RL
,
Esebua
M
,
Layfield
LJ.
Frozen section evaluation of margin status in primary squamous cell carcinomas of the head and neck: a correlation study of frozen section and final diagnoses
.
Head Neck Pathol
.
2018
;
12
(
2
):
175
180
.
4.
Hans
S
,
Delas
B
,
Gorphe
P
,
Ménard
M
,
Brasnu
D.
Transoral robotic surgery in head and neck cancer
.
Eur Ann Otorhinolaryngol Head Neck Dis
.
2012
;
129
(
1
):
32
37
.
5.
Yee
S.
Transoral robotic surgery
.
AORN J
.
2017
;
105
:
73
84
.
6.
Rinaldi
V
,
Pagani
D
,
Torretta
S
,
Pignataro
L.
Transoral robotic surgery in the management of head and neck tumours
.
Ecancermedicalscience
.
2013
;
7
:
359
.
7.
Weinstock
YE
,
Alava
I
Dierks
EJ.
Pitfalls in determining head and neck surgical margins
.
Oral Maxillofac Surg Clin North Am
.
2014
;
26
(
2
):
151
162
.
8.
Urken
ML
,
Yun
J
,
Saturno
MP
, et al.
Frozen section analysis in head and neck surgical pathology: a narrative review of the past, present, and future of intraoperative pathologic consultation
.
Oral Oncol
.
2023
;
143
:
106445
.
9.
Du
E
,
Ow
TJ
,
Lo
YT
, et al.
Refining the utility and role of Frozen section in head and neck squamous cell carcinoma resection
.
Laryngoscope
.
2016
;
126
(
8
):
1768
1775
.
10.
Hinni
ML
,
Ferlito
A
,
Brandwein-Gensler
MS
, et al.
Surgical margins in head and neck cancer: a contemporary review
.
Head Neck
.
2013
;
35
(
9
):
1362
1370
.
11.
Yu
AC
,
Afework
DD
,
Goldstein
JD
,
Abemayor
E
,
Mendelsohn
AH.
Association of intraoperative frozen section controls with improved margin assessment during transoral robotic surgery for human papillomavirus-positive oropharyngeal squamous cell carcinoma
.
JAMA Otolaryngol Head Neck Surg
.
2022
;
148
(
11
):
1029
1037
.
12.
Gwacham
NI
,
Kilowski
KA
,
Recio
FO
, et al.
Malignant peritoneal cytologic contamination with robotic hysterectomy for endometrial cancer
.
Gynecol Oncol
.
2023
;
175
:
93
96
.
13.
Gephardt
GN
,
Zarbo
RJ.
Extraneous tissue in surgical pathology: a College of American Pathologists Q-Probes study of 275 laboratories
.
Arch Pathol Lab Med
.
1996
;
120
(
11
):
1009
1014
.
14.
Zarbo
RJ.
The unsafe archaic processes of tissue pathology
.
Am J Clin Pathol
.
2022
;
158
(
1
):
4
7
.
15.
Go
M
,
Kojima
T
,
Takano
K
, et al.
Expression and function of tight junctions in the crypt epithelium of human palatine tonsils
.
J Histochem Cytochem
.
2004
;
52
(
12
):
1627
1638
.
16.
Kakarala
K
,
Faquin
WC
,
Deschler
DG.
A comparison of histopathologic margin assessment after steel scalpel, monopolar electrosurgery, and ultrasonic scalpel glossectomy in a rat model
.
Laryngoscope
.
2010
;
120
(
suppl 4
):
S155
.
17.
Troxel
DB.
Trends in pathology malpractice claims
.
Am J Surg Pathol
.
2012
;
36
(
1
):
e1
e5
.
18.
Irmakci
I
,
Nateghi
R
,
Zhou
R
, et al.
Tissue contamination challenges the credibility of machine learning models in real world digital pathology
.
Mod Pathol
.
2024
;
37
(
3
):
100422
.

Author notes

The authors have no relevant financial interest in the products or companies described in this article.

Initial findings were presented as a poster at the American Head and Neck Society (AHNS) annual meeting; July 8–12, 2023; Montreal, Quebec, Canada.