Biol Blood Marrow Transplant 25 (2019) 2281 2286 Biology of Blood and Marrow Transplantation journal homepage: www.bbmt.org Quantitative PCR Is Faster, More Objective, and More Reliable Than Immunohistochemistry for the Diagnosis of Cytomegalovirus Gastrointestinal Disease in Allogeneic Stem Cell Transplantation rez-Lledo María Sua   1,*, Maria Angeles Marcos2,3,4, Miriam Cuatrecasas3,5, Josep Antoni Bombi3,5, Francesc Fernandez-Avile s1,3,6 , Laura Magnano1, Nu  ria Martínez-Cibria n1, Noemi Llobet1, ~ ol1,3,6, Gonzalo Gutie Laura Rosin rrez-García1,3, Sofía Jorge1, Carmen Martínez1,3,6, 1,3,6  Montserrat Rovira , Alvaro Urbano-Ispizua1,3,6 1 Hematopoietic Stem Cell Transplantation Unit, Hematology Department, Hospital Clínic Barcelona, Barcelona, Spain 2 Clinical Microbiology Department, Hospital Clínic Barcelona, Barcelona, Spain 3  Biomedica August Pi I Sunyer (IDIBAPS), University of Barcelona, Barcelona, Spain Institut d’Investigacio 4 University of Barcelona, Institute for Global Health (ISGlobal), Barcelona, Spain 5 Pathology Department, Hospital Clínic Barcelona, Barcelona, Spain 6 Institut Josep Carreras, Campus Clínic, Barcelona, Spain Article history: A B S T R A C T Received 9 April 2019 Diagnosis of gastrointestinal (GI) cytomegalovirus (CMV) disease relies on the presence of GI symptoms and Accepted 12 July 2019 detection of CMV, mainly by immunohistochemistry (IHC), in GI biopsy specimens. Thus, in a symptomatic patient, a positive CMV-IHC result is accepted as a diagnosis of CMV disease. However, a positive CMV-PCR in GI Keywords: tissue is considered “possible” CMV disease. Therefore, it would be very useful if, in practice, both techniques Possible CMV disease showed equal sensitivity and reliability. This is because PCR has many practical advantages over IHC for detecting Quantitative PCR CMV. The aim of this study was to compare quantitative PCR with IHC for the diagnosis of GI CMV disease. A total Gastrointestinal tissue of 186 endoscopic GI biopsy specimens from 123 patients with GI symptoms after an allogeneic stem cell trans- Allo-SCT plantation (allo-SCT; 2004-2017) were analyzed by IHC and PCR on 113 paraffin-embedded and 73 fresh samples. The results were then compared. Of the patients with macroscopic lesions in the mucosa and CMV-IHC-positive biopsy specimens (eg, “proven” CMV disease, n = 28), all but 1 were CMV-PCR positive. Of the patients without macroscopic lesions in the mucosa and CMV-IHC-positive biopsy specimens (eg, probable CMV disease, n = 4), only 1 was CMV-PCR positive. Eight patients had CMV-IHC-negative/CMV-PCR-positive gut biopsy specimens. These cases fall within the current definition of possible CMV disease. In 6 of these 8 cases (75%), the viral load in GI tissue was very high (>10,000 copies/mg). Taken together, the results from the proven and probable cases revealed that CMV-PCR shows the same sensitivity (100%), specificity (98%), and positive (93%) and negative pre- dictive value (100%) as CMV-IHC. Detection of CMV in fresh GI mucosa by quantitative PCR is as useful as IHC for the diagnosis of GI CMV disease. The results show that quantitative PCR has the same sensitivity, specificity, and positive/negative predictive value as IHC. © 2019 American Society for Transplantation and Cellular Therapy. Published by Elsevier Inc. INTRODUCTION has decreased from approximately 35% to 8% to 10% [1]. Well- Cytomegalovirus (CMV) infection and CMV disease after known risk factors associated with CMV disease include CMV allogeneic stem cell transplantation (allo-SCT) are associated seropositivity, development of graft-versus-host disease with significant morbidity. Since the introduction of pre-emp- (GVHD), prednisone treatment, and depletion of T cells from tive anti-CMV therapy for CMV-seropositive recipients, the the graft [2-7]. The gastrointestinal (GI) system is the organ incidence of CMV disease during the first year after allo-SCT most frequently involved in CMV disease, usually the lower tract. Concomitant diagnosis of GVHD and GI CMV disease is common, but signs and symptoms of GI CMV disease (eg, diar- Financial disclosure: See Acknowledgments on page 2285. rhea, vomiting, anorexia, bleeding, and abdominal pain) are * Correspondence and reprint requests: María Sua rez-Lledo  , MD, Hematol- often indistinguishable from those of GI GVHD. Ulcers extend- ogy Department, Hospital Clínic Barcelona, Carrer Villarroel 170, 08036, Barce- lona, Spain. ing into the gut mucosa are seen upon endoscopy, but these E-mail address: msuarezl@clinic.cat (M. Suarez-Lledo ). macroscopic findings are common to both GVHD and viral https://doi.org/10.1016/j.bbmt.2019.07.016 1083-8791/© 2019 American Society for Transplantation and Cellular Therapy. Published by Elsevier Inc. 2282 rez-Lledo M. Sua  et al. / Biol Blood Marrow Transplant 25 (2019) 2281 2286 disease [8,9]. However, studies suggest that ulcers are a more Microbiological Assessment common GI CMV disease than GVHD, especially if the upper GI CMV serology (immunoglobulin G CMV) had been made in all donors and recipients before the allo-SCT. qPCR (Q-CMV Real Time Complete Kit; Nano- tract is affected [10,11]. gen Advanced Diagnostics, Buttigliera, Italy) was performed to detect CMV The difficulty in distinguishing between GVHD and CMV dis- infection in both plasma and tissue. This was carried out in the Microbiology ease in clinical practice obligates for GI CMV disease diagnosis Department. All patients were monitored weekly for CMV infection during the detection of CMV in biopsy specimens by histopathology, the first 6 months post-allo-SCT or while on immunosuppressive treatment. Patients who developed GVHD or were on steroids were monitored twice a culture, virus isolation, or immunohistochemistry (IHC) [12]. week. Pre-emptive treatment was started when CMV DNA in blood samples Detection of CMV by hematoxylin and eosin (H&E) staining of reached >1000 IU/mL. When patients were enrolled and data were analyzed, GI tissue depends on the presence of classic CMV viral inclusions. the international conversion factor for PCR-CMV was not defined. Therefore, However, viral inclusions are often not apparent; therefore, IHC analysis was based on the PCR results themselves. Data are shown with the is used by most centers to detect CMV. Thus, IHC is considered international conversion factor applied (value: PCR local laboratory £ 0.8). the gold standard for CMV detection in formalin-fixed, paraffin- Definitions (Ljungman et al CID, 2017 [20]) embedded (FFPE) tissue [13]. However, IHC can lead to an equiv- CMV infection is defined as isolation of virus or detection of viral proteins ocal interpretation due to atypical staining patterns. Therefore, (antigens) or CMV nucleic acid in any body fluid or tissue specimen. the value of IHC for detecting bona fide CMV infection depends GI CMV disease is subcategorized as follows: proven disease requires upper and/or lower GI symptoms plus macroscopic mucosal lesions (ulcers, on the observer, a fact that contributes to inter- and intralabora- sores, edema, or erythema), plus documented evidence of CMV in tissue tory variability. It is suggested that in such cases, molecular obtained by histopathology, virus isolation, rapid culture, IHC or DNA hybrid- methods (ie, quantitative PCR) could be complementary [14,15]. ization. Probable disease requires upper/lower GI symptoms plus docu- In addition, IHC for detecting inclusions in FFPE sections takes mented evidence of CMV in tissue (using the same techniques as mentioned above) but without macroscopic mucosal lesions. Possible disease requires longer (at least 48 hours) than PCR (at least 24 hours). detection of high viral DNA levels by qPCR in GI tissue (in the presence of GI Current methods used to diagnose CMV disease include symptoms), particularly when a blood sample obtained at the same time DNA hybridization and PCR. Detecting CMV using molecular does not contain CMV DNA. methods such as quantitative PCR (qPCR) is fast, highly sensi- Endoscopic Procedure tive, and more objective than IHC [16-19]. It is surprising, A multiple biopsy endoscopic study was performed using an Olympus CF- therefore, that the role of qPCR for diagnosis of CMV disease H185-L (Hamburg, Germany) endoscope. A macroscopic description of the GI remains unclear. tract was recorded. Samples were taken during the endoscopic procedure New definitions of CMV disease, which were designed origi- and sent to the pathology department for IHC (formol) and to the microbiol- nally to harmonize terminology for clinical trials, can be applied ogy department for microscopic and qPCR-CMV analysis (saline serum). to the setting of allo-SCT in cases of pneumonia and GI CMV dis- Immunohistochemistry ease. This definition establishes 3 new categories: proven, proba- FFPE samples from multiple endoscopic biopsy specimens were examined. ble, and possible CMV disease (see Methods). Of note, high viral At least 3 sections (each 5 mm thick) were processed. H&E was used for conven- DNA levels detected in tissue by qPCR in a patient showing clini- tional morphologic examination. IHC staining was performed using a monoclo- nal antibody specific for CMV (1:1 dilution; Dako, Carpinteria, CA). A result was cal symptoms are defined as possible CMV disease, particularly considered positive if more than 1 cell was detected. The median time from when a blood sample taken at the same time is negative for CMV endoscopic biopsy to diagnosis was 72 hours (range, 48 to 120 hours). [20]. Due to the lack of specific studies, there is no defined threshold for the viral load upon which a diagnosis of CMV dis- qPCR We analyzed a minimum of 5 cuts 5 mm thick for paraffin samples and a ease (either pneumonia or GI) can be based [1,20,21,22]. minimum of 10 mg of tissue for fresh samples (recommended sample size by Here, the hypothesis was that qPCR would have sensitivity, manufacturer is 10 to 40 mg). A pretreatment for paraffin samples was positive predictive value (PPV), and negative predictive value required before purification of DNA from tissue. The cuts were treated with (NPV) at least similar to that of the current gold standard (IHC) 1200 mL xylene, and after vortexing and centrifugation, the supernatant was for detection and diagnosis of GI CMV disease. We also hypothe- removed. Then, 1200 mL ethanol was added, and again after vortexing and centrifugation, the supernatant was removed. The sample preparation proce- sized that by comparing the results obtained after IHC and qPCR dure and purification of genomic DNA from paraffin-embedded tissue and of GI samples, we could define a cutoff viral load in GI tissue fresh tissue were performed with the EZ1 DNA Tissue Kit (Qiagen, Hilden, that would facilitate diagnosis and follow-up of GI CMV disease. Germany) according to the manufacturer’s specifications and using the Bio- Robot EZ1 (Qiagen). The concentration of DNA was determined by measuring the absorbance at 260 nm in a spectrophotometer. METHODS Plasma samples (500 mL) were extracted using the Qiagen Qiamp DNA General Mini Kit on a QiaSymphony system (Qiagen). We analyzed data from patients with GI symptoms who received an allo- The extraction efficiency was verified by adding an internal control to SCT and, as a result, had an endoscopic procedure performed. Biopsy speci- each sample (tissue and plasma), and an external negative control was used mens of GI tissue were examined by histology, IHC, and qPCR. In addition, to detect possible contamination. data obtained from electronic medical records were analyzed. The study was qPCR (Q-CMV Real Time Complete Kit; Nanogen Advanced Diagnostics) carried out in 2 steps. was used to detect CMV infection in plasma and tissue using a 7300 Real First, a retrospective analysis of data from patients receiving allo-SCT Time PCR System (Applied Biosystems). Limit of detection of the DNA amplifi- between 2004 and 2009 was done. IHC results at the time of diagnosis (obtained cation allows detection of the presence of 10 CMV DNA copies per PCR reac- from medical records) and PCR results obtained from paraffin-embedded sam- tion. The positive results of tissue sample were expressed as copies per ples of the same biopsy specimens stored in the institution’s biobank were microgram of DNA [23]. examined. This group is referred to hereafter as “paraffin samples.” A result was considered positive if more than 1000 copies/mg of DNA Second, data from patients receiving allo-SCT between 2010 and 2017 were detected. The median time from endoscopic procedure to diagnosis were prospectively assessed. Data were collected up until 1 year post- report was 48 hours (range, 36 to 72 hours). transplant. The results obtained by IHC and qPCR analysis of GI samples taken during the same endoscopic procedure were analyzed. Hereafter, this group Statistical Analysis is referred to as “fresh samples.” Quantitative variables were reported as the median (interquartile range) For every patient, qPCR analysis of CMV in blood samples was performed and categorical variables as absolute numbers (percentages). The sensitivity, up until 180 days post-transplant; these data were included in the analysis. specificity, and PPV/NPV of IHC and qPCR were compared. Sensitivity meas- At this institution, when an allo-SCT patient develops GI symptoms, stool cul- ures the proportion of actual positives correctly identified as such, whereas tures are taken and tested for bacteria, fungi, or viral infections (including specificity measures the proportion of actual negatives correctly identified Clostridium difficile, adenovirus, and rotavirus). If GI symptoms persist and as such. The PPV and NPV are the proportions of positive and negative results cultures are negative (results are obtained within 48-72 hours), an endo- from diagnostic tests that are truly positive and truly negative, respectively. scopic procedure is performed. The chi-square or Fisher exact test was used to compare categorical variables, rez-Lledo M. Sua  et al. / Biol Blood Marrow Transplant 25 (2019) 2281 2286 2283 and Student t test or the nonparametric Mann-Whitney U test was used to all from the lower tract (90%). In 17 of these episodes, patients assess continuous and quantitative parameters. The level of significance was presented with macroscopic lesions in the GI mucosa, and set at P < .05. All statistical analyses were performed using SPSS version 22.0 CMV was detected in endoscopic biopsy specimens by IHC [16] (SPSS, Inc, Chicago, IL). and/or inclusion bodies were detected by H&E [3]; therefore, there were 17 episodes of proven CMV GI disease. In the other RESULTS 3 episodes of CMV disease, patients had GI symptoms but no Between January 2004 and August 2017, 688 patients macroscopic lesions in the GI tract; CMV-IHC of endoscopic received an allo-SCT at the Hospital Clínic of Barcelona. The biopsy tissue was positive. These were defined as probable GI study was approved by the Ethic Committee of the Hospital CMV disease (n = 3). Clínic of Barcelona and conducted in accordance with the Decla- GVHD was diagnosed in 25 patients (40%); infections other ration of Helsinki. All patients provided informed consent. Of than CMV were diagnosed in 8 patients (13%). Other diagnoses the 688 patients, 123 (18%) fulfilled the following criteria: GI were toxicities related to mycophenolate (1.5%) and/or radio- symptoms, underwent an endoscopic procedure, and tissue therapy (1.5%) (Table 2). The remaining 14 cases (22%) had no analyzed by both IHC and PCR. The median time from allo-SCT specific diagnosis. to development of GI symptoms was 90 days (range, 45 to 270 At this time, PCR techniques on tissue were not available. days). A total of 186 GI biopsy samples were analyzed (Figure 1). For this reason, there were only cases of proven and probable Patient demographics and transplant characteristics are CMV disease. Retrospectively, CMV-PCR was performed in the shown in Table 1. paraffin samples as described in the Methods. Of the 186 samples (113 paraffin samples and 73 fresh sam- Among the proven CMV cases (n = 17), CMV-PCR was positive ples) analyzed, endoscopic examination identified 88 (47%) as in 16 (94%). However, for the 3 probable CMV cases, PCR was having macroscopic mucosal lesions (ulcers, edema, or ery- negative in all of them; these PCR results were considered false thema). Most of the samples, 149 (80%) (91 paraffin samples negatives. For the cases in which PCR of paraffin-embedded gut and 58 fresh samples), belonged to the lower tract, and only tissue was positive, the median CMV load was 1910 copies/mg 37 (20%) samples (22 paraffin samples and 15 fresh samples) (range, 1250 to 311,640 copies/mg). CMV-PCR in blood samples corresponded to the upper tract. GVHD was diagnosed in 32 taken at the same time was positive in 65% of cases. (36%); GI CMV disease in 29 (33%), 27 (94%) of the lower and 2 There were 3 additional patients (CMV-IHC negative in the (6%) of the upper tract; mycophenolate toxicity in 2 (2%); and gut) in whom CMV-PCR was positive in the endoscopic biopsy other infections (viral, bacteria, or fungi) in 18 (20%). Normal specimens. Of these, 2 had a low viral load in the tissue (1335 mucosa was identified in the remaining 98 samples (53%). Of and 3119 copies/mg). Both tissue results were interpreted as these, GVHD was diagnosed in 38 (38%); GI CMV disease in 12 false positives. One of these 3 patients had a CMV-PCR positive (12%), 11 (91%) of the lower and only 1 (9%) of the upper tract; and CMV-IHC negative (viral load in tissue 4089 copies/mg) mycophenolate toxicity in 1 (1%); and other infections (viral, but undetectable viral load in blood (defined as possible CMV bacteria, and fungi) in 7 (7%). disease according to the new definition by Ljungman et al. The diagnosis of GI CMV disease was done when GI symp- [20]). In this case, endoscopy was repeated 10 days later toms were present together with detection of CMV (by IHC) in because symptoms persisted; endoscopy identified macro- the biopsy specimens. This definition was applied in both the scopic lesions in the gut, and both CMV-IHC and CMV-PCR of paraffin and fresh samples groups. biopsy tissue were positive. Therefore, this case was finally diagnosed as proven CMV disease. Therefore, IHC showed a Paraffin Samples Group sensitivity of 95%, a specificity of 97%, a PPV of 86%, and an In total, 113 samples from 63 patients were analyzed retro- NPV of 99% for GI CMV disease. PCR in the gut biopsy speci- spectively. GI CMV disease was diagnosed in 14 patients (22%); mens showed a sensitivity of 80%, a specificity of 97%, a PPV of overall, this amounted to 20 episodes of CMV disease, almost 84%, and an NPV of 96%. N= 688 Allo-SCT 2004-2017 n= 251 Allo-SCT n= 437 Allo-SCT 2004-2009 2010-2017 Gastrointesnal Symptoms Gastrointesnal Symptoms + Endoscopic procedure + Endoscopic procedure n= 63 n= 148 Both IHC and PCR Results available Both IHC and PCR Results available n= 63 (113 samples) n= 60 (73 samples) “paraffin samples group” “fresh samples group” Figure 1. Inclusion criteria. 2284 rez-Lledo M. Sua  et al. / Biol Blood Marrow Transplant 25 (2019) 2281 2286 Table 1 macroscopic lesions in the gut was CMV-IHC positive (proba- Patients and Transplant-Related Characteristics ble CMV disease). Finally, in 8 episodes, CMV-PCR of biopsy tis- Characteristic Total Paraffin Fresh sue was positive but CMV-IHC was negative; therefore, these Patients, No. 123 63 60 were considered possible CMV disease. Of these 8 patients, 4 Male/Female, No. 72/51 36/27 36/24 (50%) had macroscopic lesions in the gut. All proven, probable, Age, median (range), y 53 (22-68) 53 (23-65) 53 (22-68) and possible cases received antiviral treatment (total of 21 epi- sodes from 13 patients). SCT type, No. (%) GVHD was diagnosed in 31 patients (52%). Other infections Sibling HLA identical 59 (48) 32 (50) 27 (45) were diagnosed in 20 (27%) cases, as well as toxicity related to Haplofamiliar 4 (3) 0 (0) 4 (6) mycophenolate (3%), Post-transplant lymphoprolipherative UNR matched 36 (29) 17 (27) 19 (32) disorder (4%), and nonspecific inflammatory features (16%) UNR mismatched 24 (20) 14 (22) 10 (17) (Table 3). Conditioning regimen, No. (%) CMV-PCR in tissue from proven and probable CMV disease Myeloablative 43 (35) 24 (38) 19 (32) cases (12 and 1, respectively) was positive in all of them Cy/TBI 31 (25) 21 (33) 10 (16) (100%), with a median viral load of 184,627 copies/mg (range, Bu/Cy 7 (5) 3 (5) 4 (7) 1250 to 5,483,875 copies/mg). Blood PCR was positive in 85% of BEAM 4 (3) 3 (5) 1 (2) these cases. There were 8 possible CMV disease cases with a Flu/Bu4 4 (3) 0 (0) 4 (7) median viral load in gut samples of 100,242 copies/mg (range, Nonmyeloablative 80 (65) 39 (62) 41 (68) 1940 to 483,261 copies/mg); of these, 62% had a positive CMV- Flu/Bu3 24 (21) 7 (11) 17 (28) PCR in blood. Of these 8 cases, 2 had a low viral load in tissue Flu/Mel 34 (27) 19 (30) 15 (25) (1940 copies/mg and 3976 copies/mg). In the first case, blood IdaFLAG 16 (13) 10 (16) 6 (10) PCR was negative and antiviral treatment was given. Of note, Flu/TBI 8 Gy 2 (2) 0 (0) 2 (3) 3 patients with mucosal ulcers and who were CMV-IHC nega- Flu/TBI 2 Gy 1 (1) 0 (0) 1 (2) tive/CMV-PCR positive (possible) underwent a repeat endo- Source of stem cells, No. (%) scopic procedure 7 days later; the samples taken then were CMV-IHC positive (proven CMV disease). Taking together all Peripheral blood 111 (91) 57 (91) 56 (93) the results, viral load in tissue was >10,000 copies/mg in 70% Bone marrow 7 (6) 4 (6) 3 (5) (9/13) of the patients with proven/probable CMV disease and Cord blood 3 (3) 2 (3) 1 (2) in 75% (6/8) of the patients with possible CMV disease. GVHD prophylaxis, No. (%) Concomitant GVHD in the gut specimen was evident in 7 CNI/MTX 37 (30) 24 (38) 13 (22) proven/probable CMV disease cases and in 5 possible CMV dis- CNI/mycophenolate 83 (68) 39 (62) 44 (73) ease cases. In 6 patients, diagnosis of both GVHD and GI CMV CNI (tacrolimus) 3 (2) 0 (0) 3 (5) disease was made at the same time (when the patient devel- ATG 24 (20) 17 (27) 7 (12) oped GI symptoms). In the remaining patients, GVHD was PT-Cy 11 (10) 0 (0) 11 (19) diagnosed previously. The median time from GVHD diagnosis CMV serostatus, No. (%) to diagnosis of GI CMV disease was 40 days. There were no sig- High risk (D /R+) 38 (31) 20 (32) 18 (30) nificant differences in the median viral load in gut specimens Intermediate risk (D+/R+) 71 (58) 34 (54) 37 (62) from proven/probable CMV disease cases with or without con- Low risk (D /R ) 14 (11) 9 (14) 5 (8) comitant GVHD (503,841 copies/mg versus 1,847,126 copies/ SCT indicates stem cell transplantation; UNR, unrelated; Cy, cyclophospha- mg, P = .314) in possible CMV disease cases with or without mide; TBI, total-body irradiation; Bu, busulfan; BEAM, BCNU (carmustine), eto- GVHD (100,242 copies/mg versus 113,565 copies/mg, P > .99). poside, cytarabine, and melphalan; Flu, fludarabine; Bu4, Busulphan 4 days; Coinfection was detected in a similar percentage of proven Bu3, Busulphan 3 days; Mel, melphalan; IdaFLAG, idarubicin, fludarabine, and possible CMV disease cases (33% versus 37%, respectively). cytarabine, and G-CSF; CNI, calcineurin inhibitor; MTX, methotrexate; ATG, antithymocyte globulin; PT-Cy, post-transplant cyclophosphamide. When analyzing proven and probable CMV disease, we found that CMV-PCR of fresh gut samples had the same sensi- tivity (100%), specificity (98%), PPV (93%), and NPV (100%) as CMV-IHC for the diagnosis of GI CMV disease. Fresh Samples Seventy-three samples from 60 patients were analyzed DISCUSSION prospectively. GI CMV disease was diagnosed in 10 patients CMV disease is not a common complication after allo-SCT, (16%), all from the lower tract. Of them, there were 12 episodes but it is associated with significant morbidity and mortality. GI where patients presented with macroscopic lesions in the gut tract problems are the most common presentation, almost mucosa and IHC detected CMV in biopsy tissue (all 12 episodes always restricted to the lower tract. For both novel and classi- were designated proven CMV disease). One patient with no cal diagnostic criteria, a patient must show clinical symptoms Table 2 Comparison of Results from IHC and PCR in GI Samples in CMV GI disease in the Retrospective Analysis of the Paraffin Samples Group Paraffin Samples (n = 113) Retrospective Microscopic Findings, No. (%) qPCR Tissue, No. (%) qPCR Blood, No. (%) GVHD Infection Other Positive Positive No CMV GI disease (n = 93) 39 (42) 14 (15) 3 (3) 3 (3) 8 (9) CMV GI disease (n = 20) 7 (35) 0 (0) 0 (0) 16 (80) 13 (65) Proven (n = 17) (symptoms, IHC+, mucosa lesions) 6 (35) 0 (0) 0 (0) 16 (95) 9 (53) Probable (n = 3) (symptoms, IHC+, normal mucosa) 1 (33) 0 (0) 0 (0) 0 (0) 3 (100) rez-Lledo M. Sua  et al. / Biol Blood Marrow Transplant 25 (2019) 2281 2286 2285 Table 3 Comparison of Results from IHC and PCR in GI Samples in CMV GI Disease in the Prospective Analysis of the Fresh Samples Group Fresh Samples (n = 73) Prospective Microscopic Findings, No. (%) qPCR Tissue, No. (%) qPCR Blood, No. (%) GVHD Infection Other Positive Positive No CMV GI disease (n = 52) 26 (50) 12 (23) 5 (9) 1 (2) 5 (10) CMV GI disease (n = 13) 7 (54) 4 (31) 0 (0) 13 (100) 11 (84) Proven (n = 12) (symptoms, IHC+, mucosa lesions) 7 (58) 4 (33) 0 (0) 12 (100) 10 (83) Probable (n = 1) (symptoms, IHC+, normal mucosa) 0 (0) 0 (0) 0 (0) 1 (100) 1 (100) Possible CMV disease (n = 8) (symptoms, IHC , PCR + in gut biopsy) 5 (62) 3 (37.5) 0 (0) 8 (100) 5 (62) Table 4 Management for Treatment in Possible GI CMV Disease Based on PCR Viral Load Tissue Viral Load Plasma Diagnosis Treatment Any >10,000 copies/mg Any Possible CMV disease CMV disease <10,000 copies/mg Any >1000 CMV infection Pre-emptive* <10,000 copies/mg Undetectable Indeterminate Wait and see* * Repeat endoscopic procedure if GI symptoms persist. for a confirmed diagnosis. However, the signs and symptoms  Any result >10,000 copies/mg in tissue could be considered affecting the upper/lower GI tract are indistinguishable from GI CMV disease, regardless of the result of PCR in blood, and those associated with GVHD. Therefore, we have to look for treated as CMV disease. other distinctive signs; this is where endoscopic biopsy can be  If viral load in tissue is <10,000 copies/mg and PCR-CMV in helpful. Some authors [10] report that macroscopic findings blood is positive, it could be considered CMV infection and such as ulcers are associated with GI CMV disease; however, pre-emptive therapy should be started. If GI symptoms per- this has not been confirmed. Here, we found that 70% of sist after 7 to 10 days of antiviral treatment, re-evaluate patients with GI CMV disease had ulcers, although ulcers are with a new endoscopy. more commonly associated with GVHD.  If viral load in tissue is <10,000 copies/mg and PCR-CMV in The next step toward a diagnosis is detection of CMV in tis- blood is negative, we would suggest a “wait-and-see” sue. IHC is the gold-standard method for histopathologic diag- approach, and if gastrointestinal symptoms persist after 7 nosis. The introduction of PCR as a diagnostic tool for CMV has to 10 days, re-evaluate with a new endoscopy. led to its incorporation for diagnosing GI CMV disease. Accord- ing to the new classification, a CMV-PCR-positive tissue is con- In summary, the data presented herein show that detection sidered possible CMV disease [20]. of CMV in fresh GI mucosa using quantitative PCR has the All diagnostic methods have limitations. The quality of the same sensitivity, specificity, PPV, and NPV as IHC for the diag- sample is the main determinant of a reliable result. PCR is a nosis of proven and probable CMV disease. Therefore, PCR is as molecular method that requires DNA extraction from the sam- useful and reliable as IHC for GI CMV disease diagnosis and is ple, followed by quantification of this DNA. Automated extrac- faster and more objective. It is important to point out that tion and quantification of viral DNA greatly reduce the there is a trend of higher viral load in proven/probable versus possibility of error and make the technique more objective possible GI CMV disease, suggesting that possible CMV disease and reproducible. By contrast, IHC requires interpretation by a could be an early diagnosis. In this line, note that there were 3 pathologist, and the results can vary depending on the thick- possible CMV disease cases that, in a second biopsy 7 days ness of the section; therefore, several sections are used to later, developed a proven CMV disease. reduce the impact of this limitation. This would be a step forward in clinical practice in patients In our experience, PCR works better for fresh samples than undergoing allo-SCT, as it would allow starting treatment of GI for paraffin-embedded samples. Formalin is not a suitable CMV disease more promptly (with a diagnosis of possible CMV medium for microbiologic tests, particularly if DNA analysis is disease) than in proven or probable CMV disease. Finally, an required. Moreover, DNA extraction is more difficult in the pres- attempt to define a cutoff for PCR-CMV in tissue to initiate ence of paraffin, a substance that also inhibits PCR reactions. antiviral treatment could be 10,000 copies/mg, in view of the Another limitation of PCR is deciding on an appropriate cut- results obtained. The qPCR could also be useful for monitoring off for the viral load. Different confounding factors may influ- the response of the treatment and to define treatment dura- ence quantification of viral DNA (eg, GVHD and/or other tion. More studies are needed to confirm these findings of infections). In our experience, the viral load during most GI CMV-PCR in tissue and to implement it in clinical practice. CMV disease episodes (proven, probable, and possible) was very high. Indeed, in 70% of cases, the blood PCR result was DECLARATION OF COMPETING INTEREST positive. There was no significant difference in median viral There are no conflicts of interest to report. load between those with GVHD and those without, bearing in mind that the number of patients analyzed in this setting was small (n = 41). ACKNOWLEDGMENTS From the data obtained herein, we defined a possible The authors thank the department of digestive endoscopy threshold for the viral load in gut specimens for the diagnosis for their efforts, dedication, and help with patient care. They of possible GI CMV disease. Because in 75% of possible CMV also thank the biobank of our institution for support and coop- disease cases, the viral load in tissue was higher than >10,000 eration. copies/mg, we suggest the following (see Table 4): Financial disclosure: The authors have nothing to disclose. 2286 rez-Lledo M. Sua  et al. / Biol Blood Marrow Transplant 25 (2019) 2281 2286 REFERENCES 13. Mills AM, Guo FP, Copland AP, Pai RK, Pinsky BA. A comparison of CMV 1. Einsele H, Mielke S, Grigoleit GU. Diagnosis and treatment of cytomegalo- detection in gastrointestinal mucosal biopsies using immunohistochemis- virus 2013. Curr Opin Hematol. 2014;21(6):470–475.. try and PCR performed on formalin-fixed, paraffin-embedded tissue. Am 2. Boeckh M, Ljungman P. How we treat cytomegalovirus in hematopoietic J Surg Pathol. 2013;37(7):995–1000. cell transplant recipients. Blood. 2009;113(23):5711–5719. 14. McCoy MH, Post K, Sen JD, et al. qPCR increases sensitivity to detect 3. Ljungman P, Brand R, Hoek J, et al. Infectious Diseases Working Party of cytomegalovirus in formalin-fixed, paraffin-embedded tissue of gastroin- the European Group for Blood and Marrow Transplantation. Donor cyto- testinal biopsies. Hum Pathol. 2014;45(1):48–53. megalovirus status influences the outcome of allogeneic stem cell trans- 15. Einsele H, Ehninger G, Hebart H, et al. Incidence of local CMV infection and plant: a study by the European Group for Blood and Marrow acute intestinal GVHD in marrow transplant recipients with severe diar- Transplantation. Clin Infect Dis. 2014;59(4):473–481. rhoea. Bone Marrow Transplant. 1994;14(6):955–963. 4. Ljungman P, Reusser P, de la Camara R, et al. Management of CMV infec- 16. Arai Y, Kondo T, Kitano T, et al. Usefulness of quantitative PCR in biopsy tions: recommendations from the infectious diseases working party of the specimens for early therapeutic intervention in gastro-intestinal cytomeg- EBMT. Bone Marrow Transplant. 2004;33(11):1075–1081. alovirus infections after allogeneic stem cell transplantation [in Japanese]. 5. Schmidt-Hieber M, Labopin M, Beelen D, et al. CMV serostatus still has an Rinsho Ketsueki. 2014;55(12):2400–2407. important prognostic impact in de novo acute leukemia patients after 17. Okahara K, Nagata N, Shimada T, et al. Colonic cytomegalovirus detection allogeneic stem cell transplantation: a report from the Acute Leukemia by mucosal PCR and antiviral therapy in ulcerative colitis. PLoS One. Working Party of EBMT. Blood. 2013;122(19):3359–3364. 2017;12:(9) e0183951. 6. El-Cheikh J, Devillier R, Crocchiolo R, et al. Impact of pretransplant donor 18. Zago  rowicz E, Przybysz A, Szlak J, Magdziak A, Wieszczy P, Mro z A. Detec- and recipient cytomegalovirus serostatus on outcome for multiple mye- tion of cytomegalovirus by immunohistochemistry of colonic biopsies and loma patients undergoing reduced intensity conditioning allogeneic stem quantitative blood polymerase chain reaction: evaluation of agreement in cell transplantation. Mediterr J Hematol Infect Dis. 2013;5:(1) e2013026. ulcerative colitis. Scand J Gastroenterol. 2018;53(4):435–441. 7. George B, Pati N, Gilroy N, et al. Pretransplant cytomegalovirus (CMV) 19. Ganzenmueller T, Henke-Gendo C, Schlue  J, Wedemeyer J, Huebner S, serostatus remains the most important determinant of CMV reactivation Heim A. Quantification of cytomegalovirus DNA levels in intestinal biop- after allogeneic hematopoietic stem cell transplantation in the era of sur- sies as a diagnostic tool for CMV intestinal disease. J Clin Virol. 2009;46 veillance and preemptive therapy. Transpl Infect Dis. 2010;12(4):322–329. (3):254–258. 8. Ljungman P, Hakki M, Boeckh M. Cytomegalovirus in hematopoietic stem cell 20. Ljungman P, Boeckh M, Hirsch HH, et al. Disease Definitions Working transplant recipients. Hematol Oncol Clin North Am. 2011;25(1):151–169. Group of the Cytomegalovirus Drug Development Forum. Definitions of 9. Cho BS, Yahng SA, Kim JH, et al. Impact of cytomegalovirus gastrointestinal cytomegalovirus infection and disease in transplant patients for use in disease on the clinical outcomes in patients with gastrointestinal graft- clinical trials. Clin Infect Dis. 2017;64(1):87–91. versus-host disease in the era of preemptive therapy. Ann Hematol. 21. Boeckh M, Stevens-Ayers T, Travi G, et al. Cytomegalovirus (CMV) DNA quan- 2013;92(4):497–504. titation in bronchoalveolar lavage fluid from hematopoietic stem cell trans- 10. Kakugawa Y, Kami M, Matsuda T, et al. Endoscopic diagnosis of cytomega- plant recipients with CMV pneumonia. J Infect Dis. 2017;215:1514–1522. lovirus gastritis after allogeneic hematopoietic stem cell transplantation. 22. Schlischewsky E, Fuehner T, Warnecke G, et al. Clinical significance of World J Gastroenterol. 2010;16(23):2907–2912. quantitative cytomegalovirus detection in bronchoalveolar lavage fluid in 11. Yin YM, Wu T, Ji SQ, et al. Cytomegalovirus enteritis after allogeneic lung transplant recipients. Transpl Infect Dis. 2013;15(1):60–69. hematopoietic stem cell transplantation [in Chinese]. Zhonghua Xue Ye 23. Yoshino T, Nakase H, Ueno S, et al. Usefulness of quantitative real-time Xue Za Zhi. 2011;32(8):516–520. PCR assay for early detection of cytomegalovirus infection in patients 12. Ljungman P, Griffiths P, Paya C. Definitions of cytomegalovirus infection and with ulcerative colitis refractory to immunosuppressive therapies. disease in transplant recipients. Clin Infect Dis. 2002;34(8):1094–1097. Inflamm Bowel Dis. 2007;13:1516–1521.