A Comparative Study of Touch Imprint and Crush Cytology with Histopathology in Upper Gastrointestinal Endoscopic Biopsies

Abstract

Background:

Touch imprint and crush cytology are simple, rapid, and cost-effective methods to distinguish between benign and malignant lesions, particularly in sentinel lymph nodes, ovarian neoplasms, and central nervous system tumors. This study was undertaken to assess the utility of imprint and crush cytology in making the diagnosis of both malignant as well as non-neoplastic diseases of the upper gastrointestinal (UGI) tract and compare the results with corresponding histopathology.

Materials and Methods:

The study was conducted on patients who underwent UGI endoscopic biopsies over a period of 1.5 years. Touch imprints and crush smears were prepared for each case and stained with Giemsa and Papanicolaou stains. The cytology results were compared with the corresponding histopathology.

Results:

The overall sensitivity and specificity of imprint and crush cytology when compared to histopathology were found to be 95.7% and 98.8%, respectively. For Helicobacter pylori infection, the sensitivity, specificity, and diagnostic accuracy of cytology were 71.4%, 99%, and 97.19%, respectively. Both crush and imprint smears had similar efficacy in rendering the cytological diagnosis.

Conclusion:

Imprint/crush smear cytology is a valuable complement to histopathology in UGI endoscopic biopsies. Evaluation of biopsy adequacy, distinguishing between malignant and benign lesions, and diagnosing specific infections such as H. pylori, Candia, and Herpes simplex virus are reliably possible on impression cytology.

INTRODUCTION

Endoscopy plays a crucial role in the diagnosis and therapy of upper gastrointestinal (UGI) lesions. Upper abdominal distress that persists despite appropriate therapy, with/without signs/symptoms suggesting organic disease, gastrointestinal bleeding, follow-up of UGI ulcers, and dysphagia are some of the most common indications, for which endoscopies are performed.[1] Endoscopic biopsy remains the gold standard for the diagnosis of UGI lesions. Imprint and crush cytology of the biopsies are rapid and cost-effective techniques which can supplement the histopathologic diagnosis. Needless to say, one of the major goals of imprint cytology is to ascertain adequacy of the biopsy and to either confirm or exclude malignancy at the earliest.[2,3,4] Data from the National Cancer Registry Programme, India, for the year 2022 showed leading five sites with the highest cancer burden in both sexes were organs of the digestive system with a cumulative risk of one in 39 persons. Nearly 37% of these occur in esophagus and stomach.[5] However, the scope of imprint and crush cytology of UGI endoscopic biopsies is not limited to malignant lesions. Among the non-neoplastic diseases, imprint cytology has proved effective in diagnosing infective etiology as well, particularly Helicobacter pylori infection.[6,7,8] This study was undertaken to assess the utility of imprint and crush cytology in making the diagnosis of both malignant as well as non-neoplastic diseases of UGI tract and compare the results with corresponding histopathology.

MATERIALS AND METHODS

A total of 107 patients who underwent UGI endoscopic biopsies (including esophagus, stomach and the second part of the duodenum) over a period of 1.5 years at a tertiary care referral center were included in the study. The research was conducted per the World Medical Association Declaration of Helsinki. Ethical approval from the institutional ethics committee was obtained. For each patient, 4–6 biopsies were taken from the lesions and touch imprints were prepared immediately by gently rolling the tissue between two glass slides. Furthermore, one of the biopsy tissues was crushed between two glass slides gently to prepare crushed smears. Four imprints and two crush smears were made, out of which three were fixed in 95% alcohol and stained with Papanicolaou and the other three were air dried, followed by fixation in methanol for Giemsa stain. The biopsies were then fixed in 10% formalin and sent for histopathological processing. On cytology, lesions were considered “unsatisfactory” when imprint slides showed low cellularity or cells obscured by blood/mucous; “negative for malignancy; reactive change/infective” when slides showed no/ mild atypia in the presence of inflammatory cells or when an organism/features of specific infection were identified; “suspicious of malignancy” when borderline atypia was seen in the presence of low cellularity; “Positive for malignancy” in presence of hypercellularity, nuclear irregularity, macro‐nucleoli, high N: C ratio, signet ring cells, mitoses, and/or necrosis. On histopathology, lesions were categorized as “inconclusive,” “non-malignant,” “dysplasia,” and “positive for malignancy.” The cytology results were compared with those of histopathology using the Chi‐square/Fisher exact test. A P value < 0.05 was considered statistically significant.

RESULTS

The study included 107 patients who underwent UGI endoscopic biopsies over a period of 1.5 years. Age ranged from 18 to 91 years (Mean age = 49 years) and M:F ratio was 2.6:1. Out of 107 endoscopic biopsies, 41 (38.3%) were from the esophagus, 40 (37.4%) were from the stomach, and 26 (24.3%) were from the duodenum. The most common presenting complaints were abdominal pain (37.4%) and dysphagia (33.6%), followed by others such as vomiting, dyspepsia, malena, hematemesis, and diarrhea. Ulceroproliferative growth (39%), ulcer (37.5%), and nodularity (30.7%) were the most common endoscopic findings in the esophagus, stomach, and duodenum, respectively. Table 1 shows the histopathological and cytological distribution of cases. Figure 1a–c show benign squamous epithelium of the esophagus, benign gastric epithelium and benign duodenal epithelium. Four cases were found to be inconclusive on histology while on cytology these were reported as “negative for malignancy” and hence were discordant. All 79 cases categorized as “non-malignant” on histology were also reported as “negative for malignancy” on cytology. Out of seven cases diagnosed as H. pylori gastritis on histology, five were correctly identified on cytology as well [Figure 1d]. In one case, H. pylori was noted on cytology smears but corresponding histology did not reveal the organism. Morphological features of Herpes Simplex virus, that is, multinucleation, molding, and margination of chromatin were identified in one case on both cytology and histology [Figure 1e]. Pseudo-hyphae of Candida were identified in the crush smears of one case [Figure 1f], while the corresponding histopathology did not reveal the organism. Out of 23 cases diagnosed as “malignant” on histology, 22 were cytologically labeled as either “suspicious for malignancy” or “positive for malignancy.” Malignancy was correctly identified on cytology in 95.6% of cases. Out of 16 cases reported as squamous cell carcinoma on histology, 10 (62.5%) were correctly typed on cytology [Figure 1g] and the remaining 6 were labeled as suspicious for malignancy; while three (75%) out of four cases of adenocarcinoma could be correctly typed on cytology [Figure 1h]. One case in esophagus was diagnosed as in situ carcinoma while on cytology it was categorized as “positive for malignancy” [Figure 2a and b]. One case of poorly cohesive carcinoma showing singly scattered atypical cells in the lamina propria on histology was misdiagnosed as non-malignant on cytology [Figure 2c and d]. For statistical analysis, cases categorized as “suspicious for malignancy” were grouped together with those diagnosed as “positive for malignancy” based on cytology findings. Table 2 presents the overall correlation between cytology and histopathology for UGI lesions. There was concordance between the two diagnostic methods in 98.1% of the cases. The sensitivity and specificity of imprint and crush cytology when compared to histopathology were found to be 95.7% and 98.8%, respectively. Positive predictive value, negative predictive value, and diagnostic accuracy in the study were found to be 95.7%, 98.8%, and 98.1%, respectively. For H. pylori infection, the sensitivity, specificity, and diagnostic accuracy of cytology were 71.4%, 99%, and 97.19%, respectively, when compared with histopathology. On comparing the imprint cytology with crush cytology, it was seen that the diagnosis was possible on both imprint and crush smears in all but two cases, one of esophageal candidiasis and one of esophageal squamous cell carcinoma, wherein the diagnosis was possible only on crush smears due to very low cellularity on imprint smears.

Table 1.

Distribution of cases on touch imprint/crush cytology with corresponding histopathological diagnosis

Cytological categories	HP diagnosis			Total
	Esophagus + GE junction	Stomach	Duodenum
Positive for malignancy	SCC-10	Adenoca-1	–	16
	SCC in situ-1	Poorly cohesive ca-2
	Adenoca-2
Suspicious for malignancy	SCC-6	Adenoca-1	–	7
Negative for malignancy	Inconclusive-4	Poorly cohesive ca-1	Lymphangiectasia-2	84
	HSV esophagitis-1	Helicobacter pylori gastritis-7	Duodenitis-8
	Esophagitis-5	Chronic gastritis-13	Brunner gland hyperplasia-4
	Barrett’s esophagus-1	Benign gastric ulcer-1	Portal HTN duodenopathy-1
	Hyperplastic polyp-1		Others (descriptive or with inflammation)- 11
	Hyperplastic polyp-1	Post peptic pyloric stricture-1
	Others (descriptive or with inflammation)-10	Others (descriptive or with inflammation)-13
Total	41	40	26	107

GE = gastroesophageal, n = number of cases, HP = histopathological, SCC = squamous cell carcinoma, Adenoca = adenocarcinoma, Ca = carcinoma, HTN = hypertensive

Figure 1.

Imprint smear from (a) esophageal biopsy shows benign squamous cells (Giemsa; ×400) (b) gastric biopsy shows benign gastric epithelium arranged in honeycomb pattern; (Giemsa, ×400) (c) duodenal biopsy shows honeycomb pattern of benign duodenal epithelial cells with interspersed goblet cells (PAP; ×400), (d) imprint smear from gastric biopsy shows numerous S-shaped Helicobacter pylori rods and benign epithelial cells (Giemsa; ×1000), (e) imprint smear from esophageal biopsy shows dysplastic looking bizarre squamous cells with multinucleation, pale nuclear inclusions (black arrow), and nuclear molding – suggestive of viral esophagitis – Herpes simplex virus (Giemsa; ×400), (f) crush smear from esophageal biopsy shows benign squamous cells and tangled fungal pseudo-hyphae of Candida (Black arrow) (Giemsa; ×1000), (g) crush smear from esophageal biopsy shows a cluster of dysplastic squamous cells – moderately differentiated squamous cell carcinoma (PAP; ×400), (h) imprint smear from gastric biopsy showing atypical gastric epithelial cells in vague gland-like arrangement – Adenocarcinoma (Giemsa; ×200).

Figure 2.

(a) Imprint smears from esophageal biopsy showing a cluster of dysplastic squamous cells – moderately differentiated squamous cell carcinoma (PAP; ×400). (b) The corresponding histopathological section shows markedly dysplastic epithelial cells, not invading the lamina propria – Carcinoma in situ (H&E; ×200). (c) Imprint smears from gastric biopsy showing benign gastric epithelium. No atypical cell noted – negative for malignancy (Giemsa; ×100). (d) The corresponding histopathological section shows benign surface gastric epithelium. However, lamina propria shows infiltration by singly scattered atypical cells – poorly cohesive carcinoma (H&E; ×200)

Table 2.

Overall cytology-histopathology correlation for upper gastrointestinal lesions

Cytology diagnosis	Histopathology diagnosis
	Malignant	Non-malignant	Total
Malignant	22 (20.6%)	1 (0.9%)	23 (21.5%)
Non-malignant	1 (0.9%)	83 (77.6%)	84 (78.5%)
Total	23 (21.5%)	84 (78.5%)	107

DISCUSSION

Imprint cytology has been employed extensively in making a preliminary distinction between benign and malignant lesions, particularly in sentinel lymph nodes, ovarian neoplasms, and central nervous system tumors. It has also been used in assessing surgical resection margins.[9] In UGI lesions, endoscopic biopsy is the gold standard for diagnosing the lesions. However, valuable time might be lost during the processing of histopathological specimens, particularly in cases of malignancies. Imprint and crush cytology, therefore, can prove to be an indispensable adjunct tool in assessing these lesions. In the present study, endoscopic biopsies from 107 patients were obtained and were subjected to imprint and crush smear preparations for cytological examination, followed by histopathological examination of the same biopsies. Malignancy was most commonly found in the age range of fifth to seventh decade which is similar to many other studies.[3,10,11] Male predominance was seen in the present study with male to female ratio of 2.6:1, which is similar to most of the study done on UGI endoscopic biopsies.[10,11,12] The spectrum of clinical presentations in the present study in gastric and esophageal lesions was similar to those seen in other studies.[10,11,12,13] Ulceroproliferative growth (39%) and ulcer (37.5%) were the most common endoscopic findings of esophageal and gastric lesions, respectively, seen in the present study which is similar to the study by Keya et al.[10]

Out of the 41 esophageal lesions, 16 cases had ulceroproliferative growth on endoscopy. All but one of these cases was diagnosed as malignant on both cytology and histopathology. Nineteen cases in the esophagus (46.3%) were diagnosed positive for malignancy and 22 cases (53.7%) were diagnosed negative for malignancy on touch imprint/crush cytology. Among these 19 malignant esophageal lesions, 18 cases showed concordance with histopathology. However, one case was diagnosed on histopathology as squamous cell carcinoma in situ (dysplastic) without any evidence of invasion. The cytology smears from this case showed malignant epithelial cells displaying moderate nuclear pleomorphism with anisonucleosis and prominence of nucleoli. This case highlights the limitation of imprint smear cytology in assessing the invasion. Batra et al.[2] reported seven cases, in which a diagnosis of esophageal malignancy was made on imprint/crush/brush cytology, but were found to be negative on histopathology.[2] They, however, did not mention the reasons for this discordance. Vijayanarsimha et al.[11] described a false positive case in their study due to the presence of regenerative atypia.[11] Among the 22 non-malignant esophageal lesions in the present study, five cases showed discordance. One case of esophageal candidiasis was diagnosed only on crush smear cytology and not on imprint smear and histopathology. The endoscopic findings of this case were also suggestive of esophageal candidiasis. This case highlights the importance of making both imprint and crush smears from different sites of the lesion. Furthermore, four other cases which had benign cytology on imprint/crush smears were considered inconclusive for opinion on histopathology because of presence of only superficial strips of squamous cells and no sub-epithelium in the biopsy. HSV esophagitis was concordantly diagnosed both on cytology and histopathology with endoscopic findings suggestive of herpes infection.

Out of the 40 gastric lesions, four cases (17.4%) were diagnosed positive for malignancy and 36 cases (42.9%) were diagnosed negative for malignancy on impression cytology. However, on histopathology, a total of five cases came out to be malignant. The discordant case was of poorly cohesive carcinoma which was reported as negative on cytology smears. Imprint and crush smears from this case showed mainly benign and reactive gastric epithelium in the background of dense mixed inflammation. No atypia or dysplasia was noted. The histopathological section from this gastric biopsy showed benign surface epithelium. However, lamina propria showed infiltration by singly scattered atypical cells. The cytohistological discordance can be attributed to the fact that with the imprint smears, there was a transfer of cells from superficial epithelium only which was benign looking in this case. Furthermore, the singly scattered atypical cells in the lamina propria might get obscured due to crushing artifacts in the crush smears. Even on re-assessing the cytological smears, these atypical cells were not identified. Similar cases have been reported in the literature wherein malignancy was missed due to misinterpretation of tumor cells in the lamina propria as benign pyloric cells or due to the exclusive subepithelial location of the tumor.[11,12] Out of the 36 non-malignant gastric lesions, six cases were diagnosed as H. pylori associated gastritis on imprint/crush cytology of which five cases showed concordance with histopathology. One case showed no H. pylori on histopathology. In histopathology, a total of seven H. pylori associated gastritis cases were diagnosed, out of which two cases showed no H. pylori on imprint/crush cytology. This may be attributed to low‑density or variable density and patchy bacterial load.[11]

Batra et al.[2] found crush smears to have a slightly lower sensitivity (89.71%) than imprint smears (94.12%) in esophageal lesions.[2] However, in the present study, cytological diagnosis was possible both on imprint and crush smears. Although, two cases had diagnostic material in crush smear slides only (one case of esophageal candidiasis and one case of esophageal squamous cell carcinoma), in the rest of the cases, crush smear examination was limited by areas of crushing artifacts.

The sensitivity, specificity, and diagnostic accuracy of H. pylori in the present study were found to be 71.4%, 99%, and 97.19%, respectively, which are somewhat comparable to those reported by Vijayanarasimha et al.[11] and are better than those reported by Arachchi et al.[7,11]. The overall sensitivity, specificity, and diagnostic accuracy of gastric imprint and crush cytology when compared to histopathology were found to be 95.7%, 98.8%, and 98.1%, respectively, which are similar to other studies.[4,10,11,12] In this study, imprint and crush cytology demonstrate a positive predictive value of 95.7% and a negative predictive value of 98.8%. The salient features of past research done on imprint and crush cytology in GI lesions is shown in Table 3. The previous studies have predominantly focused on crush cytology, which necessitates an additional biopsy sample. However, our research has demonstrated that an equally accurate diagnosis can be achieved using imprint smears, thereby eliminating the need for an additional biopsy. Although cytology can expedite the diagnostic process, it is not a substitute for histopathology. Combining both techniques can improve diagnostic accuracy and efficiency. The primary limitation of this study is the small sample size. The preparation of crush smears presents technical challenges, as excessive pressure can generate crushing artifacts, impeding the morphological assessment. However, this skill is readily learnable. In the present study, both imprint and crush smears were prepared; thus, the presence of artifacts did not influence the final interpretation of the smears.

Table 3.

Findings of previous studies done on imprint and crush cytology in GI lesions

Study	Aim	Sample size	Sites	Cytological categories	Results	Problems encountered
Singh et al.[14]	To evaluate the reliability of crush cytology by correlating cytological diagnoses with that of histopathology	89	Esophagus, GE junction, stomach, ampulla and duodenum, colon and rectum	Positive for malignancy Suspicious for malignancy Possibly reactive atypia of unknown significance Negative for malignancy	Sensitivity: 82%, Specificity: 70%, PPV: 96%, NPV: 33%, Diagnostic accuracy: 65%.	Signet ring cell carcinoma could not be identified on crush smears. Subepithelial location of tumor and well differentiated tumors were missed on cytology. Regenerative/reparative atypia was misinterpreted as malignant.
Desai et al.[3]	To evaluate the diagnostic accuracy of crush cytology for gastrointestinal lesions	451	Esophagus, GE junction, stomach, ampulla and duodenum, colon and rectum	Positive for malignancy Suspicious of malignancy Negative for malignancy	Sensitivity: 97.3%, Specificity: 90%, PPV: 99.2%, NPV: 72.5%, Diagnostic accuracy: 96.9%.	Cytology cannot substitute histopathology for tumor typing, grading, invasion confirmation. Well differentiated or low grade tumors difficult to diagnose on cytology.
Choudhary et al.[12]	To assess the accuracy of gastric biopsy imprint cytology as compared to the histopathology	79	Stomach	Positive for malignancy Suspicious of malignancy Negative for malignancy	Sensitivity: 92%, Specificity: 98.11%, PPV: 95.83%, NPV: 96.29%, Diagnostic accuracy: 94.93%	Signet ring cells and lymphoma cells were missed on imprint smears. Regenerative atypia was misdiagnosed as malignancy.
Keya et al.[10]	To assess the patterns of imprint cytology in upper gastrointestinal lesions and compare with histopathology	100	Esophagus, stomach and duodenum	Nil	Sensitivity: 98.46%, Specificity: 91.42%, PPV: 95.52%, NPV: 96.97%, Diagnostic accuracy: 96%	Subepithelial location of lesions resulted in false negative diagnosis. Dysplasia without invasion and regenerative atypia were falsely interpreted as malignant.
Vijayanarasimha et al.[11]	To correlate results of imprint cytology (including H. pylori detection) with histopathology in upper GI lesions	110	Esophagus, stomach, duodenum	Unsatisfactory Positive for malignancy Suspicious of malignancy Negative for malignancy	For malignancies – Sensitivity: 94.3% (esophagus), 88.2% (gastric), 100% (duodenum) Specificity: 100% (esophagus), 97.14% (stomach), 100% (duodenum) Diagnostic accuracy: 95.56% (esophagus), 94.23% (stomach), 100% (duodenum) For H. pylori – Sensitivity: 80% Specificity: 100% Diagnostic accuracy: 96.9%	Necrosis and subepithelial location of tumors resulted in false negative diagnosis. Regenerative atypia was the reason for false positivity.

GE = Gastroesophageal, GI = Gastrointestinal, NPV = Negative predictive value, PPV = Positive predictive value

CONCLUSION

Imprint/crush smear cytology serves as a valuable complement to the histopathological assessment of UGI endoscopic biopsies. It also aids in promptly evaluating biopsy adequacy, thus pre-empting the endoscopist for a repeat biopsy if needed at the same time. It is a rapid, cost-effective, and reliable method to distinguish between malignant and benign lesions, thereby aiding in early diagnosis and management of the patient. In addition to this, a skilled and expert cytopathologist can also make an accurate diagnosis of specific infections such as H. pylori, Candia, and Herpes simplex virus on impression cytology.

Author contributions

PG: conceptualized the work and oversaw the overall direction and planning. AK, PG, KR, and VB: contributed to the acquisition of data and its interpretation. AK: performed the statistical analysis. AK and KR: wrote the manuscript with inputs from all authors.

Ethical approval

F. No. TP (MD/MS) 19/2022/ IEC/ ABVIMS/ RMLH 863, dated July 25, 2022.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil.