Gastrointestinal Pathology in Samples From Coronavirus Disease 2019 (COVID-19)–Positive Patients

Our laboratory information system was queried for all patients who tested positive for SARS-CoV-2 reverse transcription PCR by nasopharyngeal swab and who had undergone a biopsy or resection of the GI tract between March and August 2020. Clinical data were gathered from the electronic medical record. This included information regarding patient age, sex, symptoms, indications for biopsy/resection, body mass index, comorbidities, dates of SARS-CoV-2 testing, inpatient length of stay (days), follow-up, and clinical outcomes. Endoscopic or gross findings and available laboratory values at the time of positive PCR, such as D-dimer, prothrombin time, partial thromboplastin time, white blood cell count, and anti–SARS-CoV-2 serum antibody status, were also reviewed. The project received institutional review board approval.

Gastrointestinal biopsies and resections were retrieved after diagnosis by GI pathologists (M.W., K.C., J.G., L.L.) and underwent central review by one author (M.W.). All samples were routinely processed with 10% buffered formalin. Formalin-fixed, paraffin-embedded sections were cut at 5 μm and stained with hematoxylin and eosin. Gross pathology descriptions and histology slides were reviewed. Histologic features such as ischemia, vasculitis, fibrin thrombi, active inflammation, viral inclusions, crypt apoptosis, and lymphocytosis were noted along with any other additional pathology.

SARS-CoV-2 RNA ISH was performed on 4-μm-thick formalin-fixed, paraffin-embedded sections of all specimens using RNAscope 2.5 LS probe V-nCoV2019-S (Advanced Cell Diagnostics, Hayward, California) on the fully automated Ventana Discover Ultra platform. This included colorectum (n = 13), ileum (n = 10), stomach (n = 8), duodenum (n = 6), esophagus (n = 4), liver (n = 3), pancreas (n = 1), and gallbladder (n = 1) samples. Briefly, deparaffinized sections were subjected to target retrieval for 24 minutes at 97°C in RNAScope VS Universal Target Retrieval v2 solution (Advanced Cell Diagnostics). Slides were subsequently incubated with the ready-to-use SARS-CoV-2 RNA probe mixture for 2 hours at 43°C and the signal was amplified using amplifiers (Amp1–6) according to the manufacturer's recommendation. The signal was detected using an mRNA DAB kit (Ventana, Oro Valley, Arizona). The probe targets the mRNA transcript for the S protein of SARS-CoV-2. Appropriate positive and negative RNA controls and a positive SARS-CoV-2 control were also performed.

SARS-CoV-2 was detected in GI tissue by next-generation sequencing (NGS) using a clinical assay in the James Molecular Laboratory at the The Ohio State University (Columbus). Testing was performed on 11 GI specimens from 7 patients (including all patients with positive SARS-CoV-2 nasopharyngeal PCR within a week of their procedure). This included colon (n = 5; 3 resections and 2 biopsies), terminal ileum (n = 5; 3 resections and 2 biopsies), and stomach (n = 1). RNA quality was excellent in all tissue blocks. Briefly, formalin-fixed, paraffin-embedded sections were deparaffinized and RNA extracted with the PureLink FFPE Total RNA Isolation Kit (Invitrogen, Carlsbad, California). Next-generation sequencing was performed using primer sets covering the entire SARS-CoV-2 genome (Ion AmpliSeq SARS-CoV-2 Research Panel, ThermoFisher, Waltham, Massachusetts), with reverse transcription/library preparation and sequencing performed on Ion Chef and S5, respectively (Ion Torrent, Life Technologies, Carlsbad, California). Coamplified human housekeeping gene sequences were used to ensure RNA quality. Analysis was performed in the Ion Browser with COVID-19 annotation plug-ins and strain typing was performed with a set of phylogeny tools including a local install of NextStrain (nextstrain.org).

Overall, 830 COVID-19–positive patients were discharged from our institution during the study period, including those who died. Twenty-five patients met the criteria of having nasopharyngeal swab PCR positivity for SARS-CoV-2 as well as a digestive tract–related tissue pathology specimen. Of these, 5 had positive nasopharyngeal swab PCR tests within 7 days of their biopsy or resection procedures. The remaining 20 patients had positive nasopharyngeal swab testing on an average of 53 days before their GI procedure (range, 17–121 days). The average age of the patients was 52 years (range, 19–87 years). Eleven were female and 14 were male. The average body mass index was 27.7 (range, 16.6–42.9; normal is 18.5–24.9). Comorbid conditions included obesity (body mass index >30; n = 7), diabetes (n = 8), hypertension (n = 8), history of organ transplant (n = 4), and past histories of cancer (n = 3; breast, prostate, acute leukemia). Two patients died (both 2 months after admission) and the remaining patients had an average of 5.2 months follow-up (range, 1–9 months). See supplemental digital content at https://meridian.allenpress.com/aplm in the September 2021 table of contents for additional detailed clinical and laboratory information.

Sixteen of the 25 patients (64%) experienced COVID-19–related symptoms including fever, cough, malaise, shortness of breath, or loss of taste. Of the 9 without typical COVID-19 symptoms, 7 were incidentally found to be positive on their preprocedural nasopharyngeal PCR testing and had to defer their GI procedure. The remaining 2 patients were incidentally found to be positive for SARS-CoV-2 during their admission for abdominal discomfort related to ulcerative colitis.

Gastrointestinal symptoms were documented in 16 of the 25 patients (64%). In the 5 patients with positive SARS-CoV-2 nasopharyngeal swab tests within 7 days of their procedure, GI bleeding (n = 2), ulcerative colitis flare (n = 2), and diarrhea (n = 1) were the main GI symptoms requiring gastroenterologic intervention. Three of this group also had typical COVID-19 symptoms such as shortness of breath, fever, and cough prior to their GI symptoms, whereas the 2 ulcerative colitis patients had only severe abdominal pain and an acute increase in bloody diarrhea related to their flares.

The remaining 11 of 16 patients with GI symptoms had more remote histories of SARS-CoV-2 nasopharyngeal swab PCR positivity. Their GI symptoms included nausea, vomiting, diarrhea, abdominal pain, weight loss, and blood in stool. Three of this group had only GI symptoms without other typical COVID-19 symptoms.

Pathology findings for the 5 patients with positive SARS-CoV-2 PCR tests on nasopharyngeal samples within 7 days of their GI procedure are listed in Table 1. Three had colectomies and 2 underwent biopsies.

Of the 3 bowel resections, 1 showed ischemic complications deemed secondary to COVID-19–associated coagulopathy, whereas the other 2 were likely due to severe ulcerative colitis. The ischemic case was taken from a patient who developed diarrhea, abdominal pain, and severe abdominal tenderness a few days after established COVID-19 with respiratory manifestations. A stool GI PCR panel for infectious organisms including Clostridioides difficile was negative. D-dimer levels were markedly elevated at 30.34 mg/L (reference range, <0.59 mg/L). The gross findings included a thickened sigmoid colon wall with serosal petechiae as well as necrotic-appearing splenic flexure and distal transverse colon. The mucosa was hemorrhagic, necrotic, and sharply demarcated from the background of viable tissue. Microscopic evaluation revealed diffuse mucosal necrosis with areas typical of colonic ischemia, such as attenuated, dying crypt epithelium and lamina propria hemorrhage (Figure 1, A through D). Mesenteric vessels were not thrombosed and thrombi were only seen in submucosal veins. Other areas of the bowel displayed small mucosal abscesses and features resembling pseudomembranous colitis. In situ hybridization and NGS for SARS-CoV-2 were negative.

The 2 ulcerative colitis resections were from patients with incidentally positive SARS-CoV-2 nasopharyngeal PCR tests. Both presented as having worsening ulcerative colitis flares and had been taking prednisone. No viral inclusions, increased lymphocytes, or any other features unrelated to ulcerative colitis were seen in their colectomies. The bowel resection of 1 patient showed numerous perforations and severe colitis. Pathologic findings included severe ulcerative pancolitis with deep ulcers and pseudopolyps typical of the disease. Neither ISH nor NGS identified any SARS-CoV-2 in tissue sections including small bowel and colon. The second ulcerative colitis patient had worsening abdominal symptoms after the prednisone dose was lowered. The steroid therapy had been decreased specifically because of the patient's SARS-CoV-2 result. The patient's D-dimer levels also became elevated at 1.40 mg/L. The pathologic findings, however, showed active ulcerative pancolitis with cecal sparing and inflammatory pseudopolyps typical of the disease. No distinctive vascular, thrombotic, or inflammatory findings unrelated to ulcerative colitis were seen. Although SARS-CoV-2 ISH was negative in small bowel and colon samples, NGS revealed virus in both samples. The small bowel demonstrated 290 reads and the cecum revealed 119, which is low and may be below the detection limit of ISH. It demonstrated the common G strain with the S D614G and ORF3 Q57H variants.

Both biopsy cases from patients with concurrently positive SARS-CoV-2 RNA included terminal ileum samples. One was endoscopically and histologically normal, whereas the other showed active enteritis. Interestingly, the patient with normal-appearing small bowel exhibited strong cytoplasmic and dotlike expression by ISH for SARS-CoV-2 RNA in the epithelium overlying a Peyer patch (Figure 2, A through D). The patient's accompanying colon biopsies were negative by ISH; however, both terminal ileum and colon biopsies were positive using NGS. This testing showed a common G strain with the S D614G and ORF3 Q57H variants, which was the most common virus type in the midwestern United States at the time of biopsy. The terminal ileum specimen showed a depth of 5600 reads; the right colon showed 75 reads.

The other case with biopsies exhibited active ileitis with evidence of chronic injury. The small bowel showed abnormally villous architecture with lymphocytosis, increased numbers of goblet cells lining the villi, and crypt hyperplasia. Thrombi, vasculitis, and viral inclusions were not identified. The patient had 2 separate sets of ileal biopsies, neither of which showed evidence of SARS-CoV-2 by ISH or NGS. This was despite the fact that the patient was still PCR positive on nasopharyngeal swab.

The findings of the 20 patients with COVID-19 testing beyond a week from their procedure are listed in Table 2. They had a range of GI pathologic findings, including no significant abnormalities (n = 8), active inflammation with apoptotic bodies attributed to nonsteroidal anti-inflammatory drug use (n = 1), adenomas or hyperplastic polyp (n = 4), adenocarcinoma (n = 1), graft-versus-host disease (n = 1), microscopic colitis (n = 2), erosion (n = 2), and well-differentiated neuroendocrine tumor (n = 1). All were negative for SARS-CoV-2 ISH.

One case in this group was noteworthy: virus was detected in the terminal ileum by NGS, despite the patient having a negative nasopharyngeal PCR sample. The patient's procedure was performed 17 days after having a positive SARS-CoV-2 nasopharyngeal test and COVID-19 respiratory symptoms. At the time of colonoscopy, however, the nasopharyngeal PCR sample was negative and the patient was asymptomatic for respiratory symptoms. The indication for the procedure was to identify the cause of abdominal pain of 2 months' duration. Pertinent endoscopic findings showed a cobblestoned appearance to the stomach, normal duodenum, and a single ulcer and stenosis at the ileocecal valve. Nodular mucosa was seen in the transverse colon. Microscopically, the terminal ileum and transverse colon both showed apoptotic cells involving the crypt bases and mild architectural distortion. Additionally, active enteritis with erosion and ulcer edge changes were seen in the ileum. Despite the architectural distortion, the density of chronic lamina propria inflammation was not in the range typical of a chronic idiopathic inflammatory bowel disease (IBD) diagnosis. No vascular abnormalities or viral inclusions were seen. Clinically, long-standing nonsteroidal anti-inflammatory drug use was deemed the cause for the endoscopic and pathologic findings. Although ISH studies were negative in all 4 sites, the terminal ileum was positive for SARS-CoV-2 by NGS. Sequencing gaps precluded full clade/strain typing.

SARS-CoV-2 is known to involve the GI tract. Ileal epithelium, in particular, exhibits high expression of ACE2 and TMPRSS2, which are important in the entry of this novel coronavirus into cells. The S protein of SARS-CoV-2 binds to the ACE2 receptor and is then cleaved by the host's protease TMPRSS2, exposing the S protein's subunit that allows virus and cell fusion. Gastrointestinal involvement by SARS-CoV-2 is frequently associated with symptoms of diarrhea, nausea, and vomiting. Highly persistent viral particle shedding in feces has also been documented, even after respiratory infection subsides.^9–12  One systematic review with meta-analysis found that up to 43% of COVID-19 patients were positive for SARS-CoV-2 RNA in stool specimens. Twenty-seven percent continued to show fecal positivity for viral RNA after pharyngeal swabs turned negative, for approximately a week longer.⁹  The infectivity and viability of fecal viral isolates have been questioned, however. Several studies have attempted to culture the virus in fecal samples, and few^11,12  have been reportedly successful. The ability of virus gleaned from stool samples to cause viral invasion into a cell line or result in positive nasal washes in inoculated animals has been reported, but in small numbers.^5,11  As such, although the presence of SARS-CoV-2 in GI specimens is certainly concerning for the potential for fecal-oral spread, the clinical significance of finding evidence of virus in the GI tissue itself is not completely elucidated.

In this study, our goals were to (1) elucidate the possible pathologic effects of SARS-CoV-2 on the digestive tract and (2) assess whether or not SARS-CoV-2 ISH would detect virus in patients' GI biopsies, whether concurrent with or weeks after their infection. Firstly, we described the GI specimens of 5 patients who were positive for COVID-19 during the time of their GI procedure. These patients displayed a variety of findings, but no specific histologic feature indicative of COVID-19 was identified. Moreover, the expression of SARS-CoV-2 via both ISH and NGS did not correlate with histopathologic findings. The case yielding the highest viral reads by NGS—and also the only single positive ISH result—was histologically unremarkable. The other cases reflected features indistinguishable from their underlying disease despite the GI tissues being positive for virus on NGS. Nevertheless, we found it useful to learn the effects of COVID-19–induced coagulopathy and subsequent ischemic consequences in the bowel, and to learn that we may not be able to diagnose COVID-19 in a bowel biopsy even if it is harboring virus. For our second objective, we applied ISH to all of the GI specimens, but only 1 was positive, despite the fact that multiple cases showed evidence of virus by NGS. In terms of using SARS-CoV-2 ISH for routine practice, it is useful to know that even if GI tissues are harboring virus, the viral burden may be below the detection limit of ISH. Hence, we cannot rely on a negative result.

It is challenging to explain the relationship between COVID-19 and what we found in our patients' GI tracts. Does the bowel act as a secondary site for viral entry alone or cause actual viral infection–induced pathology? In argument for the former, one report of 73 patients who tested positive for SARS-CoV-2 RNA in their stool indicated that endoscopic findings were in fact normal in infected patients. Their biopsies of GI sites such as stomach, duodenum, and rectum exhibited nonspecific changes described by the authors³  as having plasma cells, lymphocytes, and interstitial edema in the lamina propria. Our patient with a histologically normal terminal ileum that demonstrated high tissue viral reads by NGS may have represented one whose bowel was a site of viral entry without evidence of viral-induced GI disease. Although the patient was biopsied a few days before respiratory symptoms manifested, this patient never developed GI symptoms.

On the other hand, another study¹³  of 95 COVID-19 patients concluded that the presence of SARS-CoV-2 in GI tissue was indicative of severe symptoms. The study included 6 patients with worsening GI symptoms who underwent endoscopy, of whom 4 were considered to have nonsevere and 2 to have severe infection. Only the patients with severe COVID-19 had SARS-CoV-2 RNA detected in their GI tissues and/or any abnormal findings on endoscopy. A recent systematic review also supports that GI manifestations are higher in patients with severe infection requiring intensive care.¹⁴  Indeed, patient 1 in this study had a severe COVID-19 course and hypercoagulability. COVID-19 has been linked to hypercoagulability, a known cause for GI ischemia in at-risk patients.⁷  It is thought that endothelial dysfunction caused by infection results in excess thrombin generation and halting of fibrinolysis. Moreover, increased blood viscosity and hypoxia-inducible transcription factor–dependent signaling pathway may also contribute to the hypoxia in severe COVID-19.¹⁵  Hence, COVID-19 patients are at risk for ischemic bowel.⁸  Patient 1's D-dimer level was markedly elevated, which has been associated with thrombosis and worse COVID-19 outcomes in some studies.¹⁶  Although the patient did not have virus demonstrated in the bowel, the bowel pathology was a secondary consequence of the systemic effects of COVID-19 infection.

Ulcerative colitis is certainly a condition in which COVID-19 susceptibility has raised heightened concern in the gastroenterology literature. Many patients are on therapies that increase their risk of contracting certain infections. One report¹⁷  showed that age, inflammation, and location of disease were important in determining expression of ACE2 and TMPRSS2 in chronic idiopathic IBD patients. It raised the possibility that enhanced viral production and uptake in the colon can occur during active IBD given the increased expression of colonic ACE2 with inflammation. Another group of authors¹⁸  reported that IBD patients did not have higher expression of ACE2 and TMPRSS2. A meta-analysis study¹⁹  finally indicated that IBD patients are not more prone to COVID-19 infection than the general population, but that ulcerative colitis patients are overall at higher risk of hospitalization and mortality. Inflammatory bowel disease patients on steroid therapy did have higher risk of adverse COVID-19 outcomes, but biological agent use seemed protective. Both of the COVID-19–positive ulcerative colitis patients in our cohort had been taking steroids. One had evidence of virus in the colon by sequencing. Neither showed specific histologic evidence of infection or suffered COVID-19–related complications. As reports of severe IBD flares while SARS-CoV-2 positive are few, it may be useful for pathologists to know that ulcerative colitis patients may not necessarily have respiratory symptoms despite severe IBD symptoms resulting in colectomy.²⁰  As only one requisition form submitted to our pathology laboratory indicated that the patient had COVID-19, universal precautions with fresh specimen handling should be taken.

Of the group with distant histories of SARS-CoV-2 positivity, 6 had SARS-CoV-2 incidentally detected when tested specifically for their GI procedure and required postponement of the procedure for an average of 41 days. Although most of these patients were simply undergoing screening or surveillance colonoscopies, 2 had specific management deferred, including a patient with metastatic appendiceal carcinoma needing tumor debulking and a patient with a prolapsed ileostomy. Additionally, a bone marrow transplant patient with suspected graft-versus-host disease had her confirmatory endoscopic procedures deferred for 26 days after being found to be positive for SARS-CoV-2. None in this group showed evidence of virus by ISH, but we could not test all the specimens with NGS. One did have evidence of SARS-CoV-2 in the bowel specimen by NGS despite having a negative nasopharyngeal PCR test 17 days after documented infection; this has potential implications regarding risk of exposure for endoscopy staff who perform procedures on cleared patients.

In summary, GI tract specimens from COVID-19 patients in our study did not show unique histologic features of viral infection; however, it is important for pathologists to know that sequelae secondary to the systemic effects of the virus, such as ischemia, may be encountered. SARS-CoV-2 ISH detected only 1 of the cases positive for virus on NGS, and a negative result may not exclude the presence of virus in a GI specimen.