Utility of multispectral imaging in automated quantitative scoring of immunohistochemistry
- Christopher Fiore1,2,
- Dyane Bailey1,2,
- Niamh Conlon3,
- Xiaoqiu Wu1,2,
- Neil Martin4,
- Michelangelo Fiorentino1,2,5,
- Stephen Finn1,2,3,
- Katja Fall6,7,
- Swen-Olof Andersson7,
- Ove Andren7,
- Massimo Loda1,2,
- Richard Flavin1,2,3
- 1Center for Molecular Oncologic Pathology, Dana Farber Cancer Institute, Boston, Massachusetts, USA
- 2Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts, USA
- 3Department of Pathology, Trinity College, Dublin, Ireland
- 4Department of Radiation Oncology, Harvard Radiation Oncology Program, Boston, Massachusetts, USA
- 5Pathology Unit, Addarii Institute, S Orsola-Malpighi Hospital, Bologna, Italy
- 6School of Health and Medical Sciences, Örebro University, Örebro, Sweden
- 7Harvard School of Public Health, Boston, MA, USA
- 8Department of Urology, Örebro University Hospital, Örebro, Sweden
- Correspondence to Dr Richard Flavin, Department of Pathology, St James's Hospital, James's Street, Dublin 8, Ireland;
Contributors CF collected and analysed data and wrote the paper; DB and XW performed the experiments; SF, MF, NC, NM, SOA, OA and KF collected and analysed data; ML and RF conceived the idea for the paper, provided guidance and helped write the paper. ML and RF are joint senior authors.
- Accepted 6 February 2012
- Published Online First 23 March 2012
Background Automated scanning devices and image analysis software provide a means to overcome the limitations of manual semiquantitative scoring of immunohistochemistry. Common drawbacks to automated imaging systems include an inability to classify tissue type and an inability to segregate cytoplasmic and nuclear staining.
Methods Immunohistochemistry for the membranous marker α-catenin, the cytoplasmic marker stathmin and the nuclear marker Ki-67 was performed on tissue microarrays (TMA) of archival formalin-fixed paraffin-embedded tissue comprising 471 (α-catenin and stathmin) and 511 (Ki-67) cases of prostate adenocarcinoma. These TMA were quantitatively analysed using two commercially available automated image analysers, the Ariol SL-50 system and the Nuance system from CRi. Both systems use brightfield microscopy for automated, unbiased and standardised quantification of immunohistochemistry, while the Nuance system has spectral deconvolution capabilities.
Results Overall concordance between scores from both systems was excellent (r=0.90; 0.83–0.95). The software associated with the multispectral imager allowed accurate automated classification of tissue type into epithelial glandular structures and stroma, and a single-step segmentation of staining into cytoplasmic or nuclear compartments allowing independent evaluation of these areas. The Nuance system, however, was not able to distinguish reliably between tumour and non-tumour tissue. In addition, variance in the labour and time required for analysis between the two systems was also noted.
Conclusion Despite limitations, this study suggests some beneficial role for the use of a multispectral imaging system in automated analysis of immunohistochemistry.
- Cancer genetics
- cancer research
- cancer stem cells
- circulating tumour cells
- gynaecological pathology
- image analysis
- molecular oncology
- molecular pathology
- ovarian tumour
- urogenital pathology
Funding ML is supported by the Prostate Cancer Foundation, the National Cancer Institute (RO1CA131945, PO1CA89021 and P50 CA90381), the Linda and Arthur Gelb Center for Translational Research and a gift from Nuclea Biomarkers to the Jimmy Fund and the Loda laboratory. ML is also the recipient of a grant from the Dana Farber Cancer Institute–Novartis Drug Development Program.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.