A case of angioimmunoblastic T-cell lymphoma (AITL) with sufficient resected lymph node tissue and confirmed EBV infection at primary diagnosis was selected as the representative specimen for testing ISH and IHC staining conditions. A total of 15 EBV-associated malignancies were further used to validate the optimal multistaining protocol. These 15 samples included three EBV-positive DLBCL, three extranodal natural killer/T-cell lymphomas, three classical Hodgkin lymphomas (cHL) of mixed cellularity type, three AITLs, and three EBV-positive gastric carcinomas with lymphoid stroma (EBV-GCLS). Four EBV-negative malignancies including peripheral T-cell lymphoma and gastric adenocarcinoma were stained with the optimized protocol as controls. All specimens were FFPE tissues and each was cut to a thickness of 4 μm.

EBER-ISH and IHC for protein antigens were performed using the Ventana BenchMark XT automatic immunostainer following the manufacturer’s recommendations (Ventana Medical Systems, Tucson, AZ, USA). EBER-ISH was done with an ISH iView Blue Detection Kit. Protein removal and nucleic acid exposures used ISH protease 2 (#780-4148) or ISH protease 3 (#780-4149), respectively, and the reaction times were varied. Counterstaining was not performed.

Details of the primary antibodies used in IHC and probes used in ISH are summarized in Supplementary Table S1 . Two different coloring agents, 3,3'-diaminobenzidine (DAB) and new fuchsin, were used for double ISH-IHC or triple ISH-IHC staining either alone or in combination. For color development with DAB, the OptiView DAB Detection Kit (hereafter, OptiView kit) was used. Baking and deparaffinization steps were either performed or omitted. When indicated, baking was carried out at 65°C for 10 minutes. For heat-induced epitope retrieval (HIER), treatment time of cell conditioning 1 (CC1) was varied. The dilution ratio of primary antibody was varied taking into consideration the dilution ratio used in single IHC, and the reaction time was varied at 37°C. Counterstaining was performed or omitted as indicated. For color development with new fuchsin, the UltraView Universal Alkaline Phosphatase Red Detection Kit (hereafter, UltraView kit) was used. CC1 treatment time was varied. Dilution ratio and reaction times of primary antibodies were also varied. Counterstaining was performed or omitted as indicated.

EBER-ISH and IHC were also performed using another immunostainer, the Leica Bond III automated immunostainer (Leica Biosystems, Melbourne, Australia). EBER-ISH was performed with the Bond polymer refine detection kit, fluorescein-conjugated EBER oligonucleotide probe (#PB0589) and a mouse monoclonal anti-fluorescein antibody (#AR0833). Procedures were carried out in the same manner as on the Ventana equipment.

The quality of ISH-IHC staining protocols was semi-quantitatively scored by a pathologist (H.G.) and a medical laboratory scientist (J.K.J.) according to the following criteria relative to the quality of a single stain of EBER-ISH or counterstain (Supplementary Fig. S1A–S1C): score 1, no expression (<1% of cells for which target antigen expression is expected) and scarce preservation of cell or nucleus shape; score 2, rare expression (1% to <10%) and poor preservation; score 3, minimal expression (10% to <50%) and bare preservation; score 4, moderate expression (50% to <75%) and fair preservation; Score 5, abundant expression (≥75%) and good preservation. In the case of IHC, intensity of the staining and suitability of the color development position, i.e., location of expression, were evaluated from 1 to 5 points.

The quality of ISH, IHC, and hematoxylin counterstain was separately evaluated and scored. The scores were weighted based on the comparison between the results of single staining and staining according to the nominal protocol. The greater the change in the state of staining from the previous step to the later step, the greater the weight was assigned to the later step, except for the counterstain. More specifically, ISH, the first procedure of multistaining, was given a weight of 0.15 because it was minimally affected by various factors that reduce the quality of the staining. The intensity and location of IHC were given a weight of 0.30 and 0.35, respectively, because those parameters were more useful in achieving the purpose of immunostaining to evaluate the expression of a targeted antigen, while counterstaining with hematoxylin was given a weight of 0.20. The total score was added to the score of each item to which weights were applied and the maximum was 5.

Exemption from informed consent after de-identification of samples was approved by the Institutional Review Board of Asan Medical Center (2017-0522).

A simple repeat protocol according to the single staining protocols of ISH and IHC was termed the nominal protocol, P1. When EBER-ISH and CD20- or CD3-IHC staining was sequentially performed on the same slide tissue section of AITL, EBER expression was maintained at the same level as single staining, but CD20 and CD3 were poorly expressed, denaturation of the tissue was so severe that it was difficult to identify the shape of the cells and the quality of the counterstain was also poor (Supplementary Fig. S1D, S1E). As expected, EBER-ISH was not expressed at all using the reverse P1 protocol, i.e., performing IHC first and then ISH, with the same AITL case (Supplementary Fig. S2A, S2B).

EBER-ISH and CD20-IHC double staining was performed under various conditions with the same tissue used for P1. Double ISH-IHC staining was conducted to tune optimal conditions for five factors, i.e., ISH-protease, CC1, additional baking/deparaffinization, primary antibody dilution and primary antibody reaction time. These factors were considered to affect the quality of multistaining and each protocol was designated from P2-1 to DS-4 as summarized in Tables 1 and 2. The reaction time of ISH-protease 2 was varied to minimize the deterioration of CD20-IHC and counterstain and to maintain the quality of EBER-ISH. ISH-protease 3 was set to 12 minutes and ISH-protease 2 was varied to 4, 8, or 12 minutes under CC1 treatment fixed at 32 minutes. Compared with those of the single stain, quality of CD20-IHC was approximately 60% (score 0.9) and 47% (0.7) improved, respectively, and counterstain quality was the same, but the EBER-ISH quality was deteriorated (0.45) with ISH-Protease 3. Thus, ISH-Protease 3 was discarded. In protocols using ISH-protease 2, protease reaction time did not show any deterioration or enhancement in the quality of EBER-ISH, but longer reaction times of ISH-protease worsened CD20-IHC and the counterstain. In P2-2 with the shortest ISH-protease treatment, EBER-ISH quality was not deteriorated and counterstaining showed good results. However, the score for CD20 IHC was only about 42.7% (1.6/3.25) of the single stain. The total score was 3.35, the highest among P2 protocols (Fig. 1A).

CC1 reaction time (P3 protocols) was varied to 8, 16, 32, and 40 minutes. The quality of CD20-IHC staining increased with longer CC1 reaction time, but the counterstain quality gradually decreased. Among P3 protocols, the quality of staining was best for P3-2 with a total score of 3.35 (Fig. 1B). Next, the effects of baking and additional deparaffinization on the quality of staining were evaluated. Nine P4 protocols were designed by setting the baking time to 10 minutes and varying ISH-protease 2 and CC1 reaction times. In all cases, baking and deparaffinization did not deteriorate or improve the quality of staining (Fig. 1C).

Finally, the effects of dilution ratio and timing of primary antibody on the quality of multistaining were evaluated (DS protocols). Based on the optimal dilution ratio and time for CD20 single staining, the dilution ratio of CD20 antibody was halved from 1:200 to 1:100 and/or the reaction time was doubled from 60 to 120 minutes. As a result, the DS-4 protocol with a dilution factor of 1:100 of CD20 primary antibody and a reaction time of 120 minutes resulted in a total score of 5 by summing up the scores of EBER-ISH (0.75), CD20-IHC (3.25), hematoxylin (1) while maintaining a level of quality equivalent to the result of a single stain for each procedure in all evaluation factors (Fig. 1D, E).

Based on optimized conditions with EBER and CD20 staining, CD3-IHC triplicate staining was performed under various conditions with the same AITL tissue. The quality of ISH and double IHC staining was evaluated by changing the same five factors as well as the ISH-IHC staining. Among them, CC1 reaction time of CD3-IHC was varied in the same manner as CD20-IHC (Table 3). Consequentially, even when CD3 antibody was added, the best results were obtained when staining was performed in the same manner as ISH-IHC staining, which maintained the same quality of staining as single stains with each probe or primary antibody (Fig. 1F, G).

To further evaluate the applicability of the optimized protocol, ISH-IHC staining was performed on a Leica Bond III automated immunostainer with each case of AITL, cHL, and EBV-GCLS. As with a Ventana immunostainer in all three cases, the best staining quality was obtained when EBER-ISH was first applied and Enzyme I protease treatment time was minimized. In more detail, EBER was fully expressed following treatment with protease K for 10 minutes in both single EBER-ISH and ISH-IHC, and background staining occurred with protease K treatment for 20 minutes or more and when the incubation time of the EBER probe was increased from 45 minutes in single EBER-ISH to 3 hours in ISH-IHC. Antibody dilution ratio was halved and antibody reaction time was doubled compared with single immunostaining conditions (Fig. 1H, I).

The optimized multi-EBER-ISH and double IHC staining protocols were validated in 15 EBV-associated (3 cases per cancer type) and four EBV-non-associated (2 cases per cancer type) malignancies. Staining conditions were the same as in the TS-4 protocol, and the dilution ratio of each antibody was concentrated to half the dilution of the single stain while reaction time was doubled (Table 4).

As showed in Fig. 2A and B, in the EBV-positive DLBCL, (1) EBER-ISH and CD20 (red)-/CD3 (DAB)-IHC and (2) EBER-ISH and CD20 (red)-/CD30 (DAB)-IHC staining was performed to evaluate whether EBV-infected cell types were identifiable. CD20-positive cells expressing a red membrane were large with atypical nuclei and diffusely proliferated. These cells were frequently observed with dark blue color nuclei, and the surrounding small lymphocytes were admixed with red- or brown-expressing cells. These colors indicated that EBV-infected cells were not small T or B cells, but rather large atypical B cells. In addition, in EBER-ISH and CD20-/CD30-IHC staining, some EBER- and CD20-positive cells were observed to be reddish brown. When staining results were taken together, EBV-positive DLBCL were found to be neoplastic B cells infected with EBV, and some of these cells expressed CD30. Similarly, it was easy to determine which biomarker was expressed on EBV-infected cells compared to single staining in other types of EBV-associated malignancies (Fig. 2C–J, Supplementary Fig. S3). Furthermore, for all tested EBV-associated malignancies, the staining quality of each biomarker and counterstain was nearly identical to that of a single stain. In contrast, EBER was never expressed but the staining quality was well preserved in all EBV-non-associated malignancies (Supplementary Fig. S4).