To the Editor: In reply to Carvalheira et al’s1 letter regarding our article,2 we would like to answer as follows:

First, considering the urgent need for markers to distinguish between follicular adenoma (FTA) and carcinoma (FTC) during FNA cytology, we find it remarkable that no clinical studies have been published validating the four classifiers originally described by Cerutti et al.3 Indeed, we are unaware of any diagnostic laboratory using these markers even though they were originally estimated to have a sensitivity of 1.00 for detecting malignancy.

In their letter, Carvalheira et al.1 express a concern regarding the specificity of the custom polyclonal antibodies we raised against FAM129A and STT3A. Readers of Modern Pathology, and especially those who use IHC techniques in the diagnostic laboratory, are well aware of the thorny issues that can arise when assessing antibody specificity. In this regard, two of us (DJW and EP) are participants in the International Society of Oncology and Biomarkers (ISOBM) TD-workshops, which since 1996 have characterized the epitope structure and specificities of >350 diagnostically relevant monoclonal antibodies.4

Our anti-FAM129A and -STT3A were raised using synthetic peptides identical to those used by Cerutti et al.3 Although results from peptide-blocking studies must be evaluated with caution,5 the ability of unconjugated peptide to abrogate tissue staining strongly supported the anti-peptide specificity of our reagents. It was during western blotting experiments (to assess off-target binding) that we found significant expression of ARG2, STT3A and FAM129A in many FTA lysates. Cerutti et al.3 also characterized their anti-FAM129A and anti-STT3A by western blotting. However, although they included lysates from papillary thyroid carcinoma, FTC and normal thyroid, they failed to include any FTA tissue. This seems a remarkable omission considering they were developing reagents to discriminate FTA from FTC.

In their letter, Carvalheira et al. suggest that the absence of a 70-kD band in our FAM129A western blots indicates that our antiserum has a different reactivity to the reagent they used. We refer them to Figure 5 where a band of this size can be seen in lanes 1, 2 and 7. In agreement with their observations, we found some tissues (lane 8) that only display the 70-kD FAM129A fragment. If the 70-kD band is indeed a ‘stress marker’,6 we probably failed to detect it in the FTC-133 cell line simply because we prepare lysates from log-phase cells. Do Carvalheira et al. use confluent ‘stressed’ cultures?

In regard to the somewhat smaller band seen in our anti-STT3A western blots, discrepancies between predicted and measured protein sizes frequently occur. A plethora of factors acting alone, or together, can influence protein mobility in SDS-PAGE. Furthermore, during western transfers small errors in the registration of gel/membrane/film can occasionally occur. A quick web-based search for STT3A western blots show reported reactivity from around 60 kD to as high as 90 kD.

In their letter, Carvalheira et al.1 discuss the specificity of our ‘in-house’-generated reagents but avoid mention of the commercial anti-ARG2 and GADD153 antibodies they used. In their original paper, no statement was made about how they validated these commercial reagents. This is surprising and raises some concern since we and others7 have found the anti-GADD153 reagent to show very poor selectivity.

In our article we clearly state that we performed extensive staining trials including an assessment of the endogenous peroxidase blocking specifically mentioned by Carvalheira et al.1 As the principal aim of the method is to discriminate FTC from FTA, we also performed parallel antibody titrations on serial paraffin sections of both tumor types. In our laboratory, internationally standardized fixatives, time prior to fixation, tissue processing and length of fixation time are used. We handle tissue samples from most organs, utilizing hundreds of commercial antibodies in everyday diagnostics as well as in research projects. New antibodies are introduced into our routine on a regular basis, with a standardized way of implementation. Cases from our laboratory are regularly sent for second opinion to approved laboratories in different countries. These immunostained slides are seldom questioned. We therefore find it highly unlikely that factors concerning fixation/antibody dilution/blocking would be of any significance. Indeed, it is generally agreed that all markers for use in diagnostics ought to be robust. A marker is of no practical use if even small differences in, for example, fixation time will have a significant effect on the result.

In conclusion, our peptide-blocking, western and IHC data clearly demonstrate that our custom reagents show good specificity and that STT3A, ARG2 and FAM129A are expressed in a significant number of follicular adenomas. Data from western blotting and IHC studies are concordant even though the tissues are processed in a very different manner prior to analysis.

Thus, we believe that the suggestions made by Carvalheira et al. do not abrogate the conclusion that DDIT3, STT3A, ARG2 and FAM129A are not clinically useful in distinguishing between follicular thyroid adenoma and follicular thyroid carcinoma.