Morphometric characterization of collagen and fat in normal ventricular myocardium

Highlights • Automated image analysis is a useful tool for cardiac tissue quantification.• Collagen and fat proportions are demonstrably higher in the right ventricle.• We present reference values for collagen and fat proportions in normal myocardium.


Introduction
Digital quantification of myocardial tissue structure has emerged as an important tool in the pathological and clinical evaluation of cardiovascular disease. Excessive accumulation of collagen within the extracellular matrix, the hallmark of myocardial fibrosis, results in pathological structural remodeling of the heart muscle. Recently, histomorphometric analysis has been employed in the quantification of tissue components and fibrosis in arrhythmogenic cardiomyopathy [1] and Brugada syndrome [2] . Semiautomated histological quantification of fibrosis has also been used to validate findings from cardiac magnetic resonance T1 mapping in patients with severe aortic stenosis [3] . To define pathology, knowledge of the relative proportion of collagen, fat, and cardiomyocytes in normal ventricular myocardium is necessary. The aim of the study was to determine histological cardiac tissue composition and collagen content using automated digital pathology software.

Material and methods
We identified 29 noncardiac death cases with retained cardiac tissue referred to our national cardiac pathology center. Ethical approval for this study was obtained from the UK National Health Service Research Ethics Committee (ref. 10/H0724/38) and conformed to the principles outlined in the Declaration of Helsinki. Informed consent was provided by the next of kin at the time of referral. Demographics, clinical data, toxicological, and pathological findings were entered prospectively into a database. were macroscopically and microscopically normal following examination by expert cardiac pathologists. Tissue sections (5-μm thickness) were sampled at the mid-ventricular level or from the anterior aspect of the right ventricular outflow tract. In total, 6 regions were evaluated: right ventricular outflow tract; right ventricle (RV); anterior interventricular septum (IVS); posterior IVS; anterior left ventricle (LV); and posterior LV. The RV section included 3 sampled locations from the right anterior, lateral, and posterior walls. Right ventricular tissue areas were combined for the purpose of morphometric quantification. Sections were stained with Picrosirius Red (PSR) for collagen ( Fig. 1 ) and slides were scanned using 20 × magnification on an automated high-resolution scanner (Hamamatsu Nanozoomer). Automated calculation of crosssectional tissue area and quantification of collagen (%), myocytes (%), and fat (%) was performed using an application developed within Visiopharm image analysis software (Visiopharm A/S, Hoersholm, Denmark). Perivascular collagen surrounding epicardial and intramural vessels was excluded due to high variability in vessel caliber between samples. Exclusion was based on automated lumen detection (elliptical score ≥0.4) incorporating surrounding region of PSR stain, as shown in Fig. 2 . Total tissue area was categorized into epicardial and endocardial regions (50:50). Continuous data are summarized as means ± standard deviations and categorical data as groups' percentages. Proportions (log transformed) of each tissue component were analyzed according to sampling location (LV, IVS, and RV) using a multilevel model, accounting for the clustered structure of the data. Predictions at the original scale were expressed in terms of geometric means across groups in the data.

Results and discussion
The majority of decedents were male (25/29; 86%). The mean age at death was 32.1 ± 9.9 (range 18-54) and mean BMI 28.7   Table 1 outlines predicted values for the proportion of collagen, fat, and myocytes (total, epicardial, and endocardial) in the RV, IVS, and LV. There was insufficient evidence to support differences in collagen ( P = .08) or fat ( P = .06) proportion according to sex; predicted values for males (n = 25) and females (n = 4) are shown in Table 2 .
Our results demonstrated that RV collagen and fat proportions were significantly higher than the LV, particularly within the RV epicardium (15.9% and 20.1%, respectively). This is in line with earlier observations reporting higher concentrations of collagen within the right heart [4] . While stereological techniques are recognized as the gold standard in collagen quantification [5] , we used a morphometric and automated approach that included anal-  ysis of whole tissue sections. Our data did not yield sex-specific differences in collagen and fat proportion, which may be due to the small number of females included in our study. This is reflected in the respective CIs shown in Table 2 .
Overall, our study highlights the spectrum of normal variation of collagen and fat content within the RV, in contrast to the pathological changes seen in diseases such as arrhythmogenic cardiomyopathy [6] . Although we have previously reported lower ventricular collagen proportions in a control group of 6 structurally normal hearts [2] , this analysis was performed using semiautomated color thresholds. Lower proportions may reflect the imprecise nature of semiautomated approaches to collagen quantification, which can underestimate collagen area and affect reproducibility. Indeed, a recent study reported an approximate 50% reduction of detected collagen area using a semiautomated approach when compared with automated or stereological light microscopy techniques [7] . Moreover, the effect of tissue thickness on collagen quantification should not be understated. In the aforementioned study, there was a positive relationship between automated collagen quantification and increasing section thickness. This underscores the importance of using thin sections of paraffin-embedded tissue and maintaining consistency across samples. Our data highlight the potential utility of automated histological quantification to complement expertise provided by the pathologist and should be interpreted in the context of thin tissue sections and comparable sampling location.

Conclusions
This is the first study to report regional proportions of cardiac tissue composition in normal human ventricular myocardium. Previous work has demonstrated the validity of automated methods of histological image quantification in the assessment of myocar-dial fibrosis [7] . Here, we provide location and sex-specific proportions that represent reference values to aid quantitative evaluation of ostensibly diseased samples in future studies.

Author contribution
C.M., J.W., P.B., and A.J. were involved in the conception, design, and drafting of the article. C.M., A .A ., M.P., S.S., E.R.B., and M.N.S. were involved in the interpretation of data and critical revision of content. I.C.S. supervised all the statistical aspects of this study including design; analyses were conducted by C.M. All authors have read and approved the final article.