In this study, we have compared and validated commercial pan-Trk clones using different assays in order to properly identify NTRK-rearranged tumors. We identified clones and assays/protocols that can be reliably used by pathologist for the screening of NTRK rearranged tumors in a diagnostic routine setting. IHC is considered a screening method, therefore the sensitivity of the assays used, is crucial. Cases with a negative IHC result, likely will not undergo further NTRK fusion testing. In the here presented study two antibody clones performed very well in terms of sensitivity, EPR17341 (100% independent on protocol) and A7H6R (93.8%), missing maximum one out of 16 cases with a pathogenic NTRK fusion. These results are slightly higher compared to the literature, as a recent metanalysis showed an overall sensitivity of 82% in a total of 224 NTRK-rearranged solid tumours14. The reason for this might be that the awareness and experience among pathologist improved over time in using and scoring pan-Trk immunohistochemistry assays. There are several webinars and publications available that increase the awareness and train skills to identify weak and/or focal immunohistochemical staining in NTRK fusion-positive tumors, this is especially true for NTRK3-rearranged tumors 11,18,23,24 and might be also true for certain tumor types25. To avoid any bias due to weak or partial staining, only whole slide sections were used in this study.
It is also important to understand that neoplasms showing neuronal, or muscle differentiation may often show corresponding wildtype Trk expression 26,27. Therefore, it is not surprising that sensitivity and specificity of pan-Trk IHC, to detect fusion driven sarcomas, has been considered one of the lowest among different tumor types18. For this reason, a BCOR-rearranged subgroup of mesenchymal tumors was added in our molecularly confirmed NTRK fusion-negative cohort and needs to be highlighted. It has been described that BCOR rearrangements often lead to an NTRK3 upregulation on both, mRNA and protein levels 28. Although the number of analyzed cases of this rare entity was low (n = 5), great differences in specificity were seen with 60% and 40% for EPR17341 and EP1058Y respectively, but 100% for A7H6R. This leads to the speculation that certain antibody clones could be preferably used for specific organ systems or tumor types. On the other hand, no “organ-specific” clone could so far be identified for example in the salivary gland subgroup (overall specificity of about 54%). Number examined was low, whether this holds true and is also true for other tumor types needs to be further investigated.
Specificity of pan-Trk IHC has also been previously analyzed in broader sequencing studies and in a meta-analyses14,18. In our selected wild type cases, we included cases that potentially cause false positive results to better identify the quality among different antibody clones and protocols included in this study. As positive cases have been described in NTRK amplified tumors, this molecular subgroup was also explored in our study 17. A correlation between the copy number (copy number gain) and positive staining in our study was only seen in EPR17341 when used the in-house protocol (ŋ2=13.8%) but not in the others, concluding that the number analyzed is low and therefore no statement can confidentially be made. When looked at positive cases in an organ/entity-based fashion various breast carcinomas showed positivity. This is not a novel finding since poor sensitivity in breast carcinomas has previously been described by others15,18.
All the factors noted above have to be taken into account, while screening is performed for these tumors, and a proper molecular testing has to be undertaken in cases with positive staining. Furthermore, if NGS testing is available and routinely done, pan-Tkr IHC should be considered for validation, to confirm a transcribed and translated fusion protein that is active. This study nicely showed that two non-pathogenic NTRK – rearrangements did not stain positive in pan-TRK immunohistochemistry. This might become especially relevant for non-activating NTRK fusions, mainly detected on a DNA level. According to our study, the clones EPR17341 and A7H6R can be used for such an approach as both of our cases with a non-activating NTRK fusion stained negative.
Only a limited number of studies have compared different pan-TRK clones and assays, with molecularly confirmed NTRK-rearranged tumors (n = 1 and n = 0 respectively) 15,16. This is mainly due to the lack of NTRK-rearranged tumor tissue availability. We here presented the result of 18 NTRK-rearranged tumors with all 3 NTRK genes involved in rearrangements. The Clone EPR17341 has so far been the most widely reported and used antibody, available as a class I in vitro diagnostic analytic assay using Benchmark IHC/ISH instruments from Ventana-Roche 22 and has been evaluated in this study together with A7H6R (Cell signaling) and EP1058Y (Abcam). Rearrangement specific staining patterns have been confirmed and from the other, here tested antibodies, each assay shows its own sensitivity and specificity score. For EPR17341, A7H6R and EP1058Y clones, an in-house protocol was established. For EPR17341 (in house protocol), only small differences to the RTU (Ventana, Roche) protocol were identified. These differences could be dependent on dilution, as the RTU assay comes prediluted, incubation and/or pretreatment time, making careful establishment of any clone important. We conclude that given the rarity of NTRK fusions but the therapeutic importance, testing for these rearrangement in daily practice has become necessary. As access to molecular tests including NTRK1-3 can be limited to bigger center hospitals, IHC can offer a fast and affordable screening tool in any practice. Based on our data both clones EPR17341 and A7H6R show a high overall sensitivity (> 94%) and an overall specificity above 72% and can therefore be recommended for clinical use. Nevertheless, a careful protocol establishment and the inclusion of on-slide controls are critical and recommended. This study does not take therapy and treatment responses into account. Therefore, this study is not a clinical validation study, and no conclusions can be drawn on the predictive value of pan-Trk for targeted therapy and on the validity of the chosen cutoffs.