As the number of immunostains increase on a slide, particularly when by design they overlap, co-localization and accurate interpretation becomes more difficult. From a laboratory perspective, these “cocktails” can be difficult to design given a limited number of wavelength options, variable staining intensity and inevitable background staining. These limitations in combination make interpretation of such slides difficult and time consuming with the naked eye. Multispectral image analysis is one purported solution for this problem that has been recently studied in fine needle aspiration of the thyroid[2
] and examined in a review of quantitative imaging techniques.[3
The staining results presented here are consistent with past studies at the University of Pittsburgh Medical Center performed with an equivalent cocktail.[4
] They support patterns of p53 and CK20 expression that vary by diagnosis as reported previously in the literature.[5
] Subjectively, the addition of multispectral imaging permitted much easier interpretation aided by enabling the manipulation of the staining patterns. The use of programmable macros greatly improved the efficiency of capturing that information accurately and quickly. Built into the Nuance software are tools, not presented here, that permit objective quantification of staining density and co-expression, worthy of future investigation. Future improvements not limited to enhanced spectral density, accommodating larger number of immunostains, enabled by combining fluorescence mode in combination with quantum-dots[7
] may provide a level of sophistication to immunohistochemistry not yet seen and ripe for future investigation.
As this study was a pilot, there are a number of important limitations worthy of discussion. CD44 expression was poor in the cocktail batch; an unanticipated result given the success with it in previous studies. In selected cases, bright-field analysis was able to compensate to a sufficient extent to reliably discriminate the three stains despite the spectral similarity between the CD44 and P53 chromogens. This was the exception rather than the norm, and therefore, felt to represent anecdotal evidence of Nuance’s ability to manipulate 3 stains under bright-field examination, but requires further scrutiny. Even with appropriate staining, the spectral signatures of the chromogens used with the CD44 and P53 antibodies were overly broad with considerable overlap. Quantum dots have been shown to provide much narrower spectral signatures and would make an interesting comparison. Further investigation is also needed to compare the results from Nuance’s multispectral camera to a “spectral unmixing” RGB method, as described by Ruifrok et al
Lastly, the value of compressing multiple slides into a single tissue section was of limited benefit. With rare exception, extra blank slides cut with the first facing of the block would have sufficed. However, one situation where this value may be realized is an unexpected finding on a cytologically obtained cell block, a setting that invites further investigation.
In conclusion, the combination of multispectral microscopy and multiple immunostain cocktails form a promising tool for the interpretation of cases with limited material such as urothelial needle core biopsies. This is true moreover for applications that favor immunohistochemical stains without the need for fluorescent markers or the expense of confocal microscopy. The fields of cytopathology, hematopathology and surgical pathology will serve as prominent subjects for further experimentation in the future.