Computer-assisted mitotic count using a deep learning–based algorithm improves interobserver reproducibility and accuracy

The mitotic count (MC) is an important histological parameter for prognostication of malignant neoplasms. However, it has inter- and intraobserver discrepancies due to difficulties in selecting the region of interest (MC-ROI) and in identifying or classifying mitotic figures (MFs). Recent progress in the field of artificial intelligence has allowed the development of high-performance algorithms that may improve standardization of the MC. As algorithmic predictions are not flawless, computer-assisted review by pathologists may ensure reliability. In the present study, we compared partial (MC-ROI preselection) and full (additional visualization of MF candidates and display of algorithmic confidence values) computer-assisted MC analysis to the routine (unaided) MC analysis by 23 pathologists for whole-slide images of 50 canine cutaneous mast cell tumors (ccMCTs). Algorithmic predictions aimed to assist pathologists in detecting mitotic hotspot locations, reducing omission of MFs, and improving classification against imposters. The interobserver consistency for the MC significantly increased with computer assistance (interobserver correlation coefficient, ICC = 0.92) compared to the unaided approach (ICC = 0.70). Classification into prognostic stratifications had a higher accuracy with computer assistance. The algorithmically preselected hotspot MC-ROIs had a consistently higher MCs than the manually selected MC-ROIs. Compared to a ground truth (developed with immunohistochemistry for phosphohistone H3), pathologist performance in detecting individual MF was augmented when using computer assistance (F1-score of 0.68 increased to 0.79) with a reduction in false negatives by 38%. The results of this study demonstrate that computer assistance may lead to more reproducible and accurate MCs in ccMCTs.

Phosphohistone H3 Immunohistochemical Labeling: a Potentially Useful Tool for Risk Stratification and Prognostic Analysis in Gastrointestinal Stromal Tumors.

BackgroundIt is well recognized that risk stratification of gastrointestinal stromal tumors (GISTs) is closely related to tumor size, mitotic index (MI), and primary location. Among these three parameters, tumor size and primary location are easily established, while MI is subjective and its repeatability is poor. It is thus necessary to identify a biomarker to represent the true MI. Expression status and biological or prognostic significance of mitotic marker phosphohistone H3 (PHH3) and cell proliferation marker Ki67 in GIST have not been clearly understood until now. MethodsAn immunohistochemistry experiment was performed to detect the expression status of PHH3 and Ki67 in 125 paraffin-embedded GIST samples. All of the patients were followed up until September 30, 2019. ResultsThe MI determined using stained hematoxylin and eosin (H&E) sections (MI-H&E) and immunohistochemically positive PHH3 index (PHH3-IHC) was compared among groups of different genders, ages, primary locations, and histological subtypes, showing that the difference was not statistically significant (P>0.05). MI-H&E and the immunohistochemically positive Ki67 index were positively correlated (r=0.273, P=0.001), but the correlation was lower than that with the PHH3-positive index (r=0.705, P=0.000). The PHH3-positive index was also positively correlated with the Ki67 index (r=0.224, P=0.006). MI-H&E were positively correlated with MI quantified using immunohistochemically stained PHH3 sections (MI-PHH3) (P<0.05). After using PHH3 to perform MI quantification, the risk stratification of five GIST cases was changed, where two cases were given a higher risk grade and three cases were given a lower risk grade. Follow-up data were obtained from 98 cases (98/125, 78.4%), including two cases of metastasis and one death. Both metastatic and death cases had high MI-H&E. One metastatic case and one death case had high PHH3-positive indexes, while the remaining metastatic case had a low PHH3-positive index. ConclusionImmunohistochemical PHH3 labeling is a potentially useful tool for risk stratification and prognostic analysis in GIST. Using immunohistochemical PHH3 labeling makes it more convenient for pathologists to determine the MI for GIST. MI quantification with immunohistochemical PHH3 sections can be used as an adjunct tool for risk stratification and prognostic analysis of GIST, but cannot completely replace MI quantification using stained H&E sections. The Ki67 index is positively correlated with MI-H&E, although the efficiency of tumor risk stratification is lower than that of PHH3.

Computer-Assisted Mitotic Count Using a Deep Learning-based Algorithm Improves Inter-Observer Reproducibility and Accuracy in canine cutaneous mast cell tumors

The mitotic count (MC) is an important histological parameter for prognostication of malignant neoplasms. However, it has inter- and intra-observer discrepancies due to difficulties in selecting the region of interest (MC-ROI) and in identifying/classifying mitotic figures (MFs). Recent progress in the field of artificial intelligence has allowed the development of high-performance algorithms that may improve standardization of the MC. As algorithmic predictions are not flawless, the computer-assisted review by pathologists may ensure reliability. In the present study we have compared partial (MC-ROI preselection) and full (additional visualization of MF candidate proposal and display of algorithmic confidence values) computer-assisted MC analysis to the routine (unaided) MC analysis by 23 pathologists for whole slide images of 50 canine cutaneous mast cell tumors (ccMCTs). Algorithmic predictions aimed to assist pathologists in detecting mitotic hotspot locations, reducing omission of MF and improving classification against imposters. The inter-observer consistency for the MC significantly increased with computer assistance (interobserver correlation coefficient, ICC = 0.92) compared to the unaided approach (ICC = 0.70). Classification into prognostic stratifications had a higher accuracy with computer assistance. The algorithmically preselected MC-ROIs had a consistently higher MCs than the manually selected MC-ROIs. Compared to a ground truth (developed with immunohistochemistry for phosphohistone H3), pathologist performance in detecting individual MF was augmented when using computer assistance (F1-score of 0.68 increased to 0.79) with a reduction in false negatives by 38%. The results of this study prove that computer assistance may lead to a more reproducible and accurate MCs in ccMCTs.

Mechanical Unloading is Associated with Decreased DNA Content in Cardiomyocytes Independent of Nucleation State

ABSTRACTBackgroundCardiomyocytes increase DNA content in response to stress in humans. Proliferation has been reported in cardiomyocytes in failing hearts and following LVAD unloading which may represent a resolution of this process through cell division. However, cardiac recovery from LVAD is rare.MethodsWe quantified cardiomyocyte nuclear number, cell size, DNA content and the frequency of cell cycling markers by imaging flow cytometry from human subjects undergoing LVAD implantation or primary transplantation.ResultsCardiomyocyte size was 15 percent smaller in unloaded versus loaded samples without differences in the percentage of mono-, bi, or multi-nuclear cells. DNA content per nucleus was significantly decreased in unloaded hearts versus loaded controls. Cell cycle markers, Ki67 and phosphohistone H3 (H3P) were not increased in unloaded versus failing samples.ConclusionsUnloading of failing hearts is associated with decreased DNA content of nuclei independent of nucleation state within the cell. As these changes were associated with a trend to decreased cell size but not increased cell cycle markers, they may represent a regression of hypertrophic nuclear remodeling and not proliferation.

Mitotic and Proliferative Indices in WHO Grade III Meningioma

Meningiomas with inherently high mitotic indices and poor prognosis, such as WHO grade III meningiomas, have not been investigated separately to establish interchangeability between conventional mitotic index counted on H&E stained slides (MI) and mitotic index counted on phosphohistone-H3 stained slides (PHH3 MI). This study investigates the agreement of MI and PHH3 MI and to analyze the association of progression-free survival (PFS) and MI, PHH3 MI, and the proliferative index (PI, Ki-67) in WHO grade III meningioma. Tumor specimens from 24 consecutive patients were analyzed for expression of Ki-67, PHH3 MI, and MI. Quantification was performed independently by two observers who made replicate counts in hot spots and overall tumor staining. Repeatability in replicate counts from MI and PHH3 MI was low in both observers. Consequently, we could not report the agreement. MI, PHH3 MI and hot spot counts of Ki-67 were associated with PFS (MI hot spot HR = 1.61, 95% CI 1.12–2.31, p = 0.010; PHH3 MI hot spot HR = 1.59, 95% CI 1.15–2.21, p = 0.006; Ki-67 hot spot HR = 1.06, 95% CI 1.02–1.11. p = 0.004). We found markedly low repeatability of manually counted MI and PHH3 MI in WHO grade III meningioma, and we could not conclude that the two methods agreed. Subsequently, quantification with better repeatability should be sought. All three biomarkers were associated with PFS.

Myh6-driven Cre recombinase activates the DNA damage response and the cell cycle in the myocardium in the absence of loxP sites

ABSTRACTRegeneration of muscle in the damaged myocardium is a major objective of cardiovascular research, for which purpose many investigators utilize mice containing transgenes encoding Cre recombinase to recombine loxP-flanked target genes. An unfortunate side effect of the Cre-loxP model is the propensity of Cre recombinase to inflict off-target DNA damage, which has been documented in various eukaryotic cell types including cardiomyocytes (CMs). In the heart, reported effects of Cre recombinase include contractile dysfunction, fibrosis, cellular infiltration and induction of the DNA damage response (DDR). During experiments on adult mice containing a widely used Myh6-merCremer transgene, the protein product of which is activated by tamoxifen, we observed large, transient, off-target effects of merCremer, some of which have not previously been reported. On Day 3 after the first of three daily tamoxifen injections, immunofluorescent microscopy of heart sections revealed that the presence of merCremer protein in myonuclei was nearly uniform, thereafter diminishing to near extinction by Day 6; during this time, cardiac function was depressed as determined by echocardiography. On Day 5, peaks of apoptosis and expression of DDR-regulatory genes were observed, highlighted by >25-fold increased expression of Brca1. Concomitantly, the expression of genes encoding cyclin-A2, cyclin-B2 and cyclin-dependent kinase 1, which regulate the G2/S cell-cycle transition, were dramatically increased (>50- to 100-fold). Importantly, immunofluorescent staining revealed that this was accompanied by peaks in Ki67, 5′-bromodeoxyuridine and phosphohistone H3 labeling in non-CMs, as well as CMs. We further document that tamoxifen-induced activation of merCremer exacerbates cardiac dysfunction following myocardial infarction. These findings, when considered in the context of previous reports, indicate that the presence of merCremer in the nucleus induces DNA damage and unscheduled cell-cycle activation. Although these effects are transient, the inclusion of appropriate controls, coupled with an awareness of the defects caused by Cre recombinase, are required to avoid misinterpreting results when using Cre-loxP models for cardiac regeneration studies.This article has an associated First Person interview with the first author of the paper.

Improving precise counting of mitotic cells in mantle cell lymphoma using phosphohistone H3 (PHH3) antibody

AimsMantle cell lymphoma (MCL) has a highly heterogeneous clinical course ranging from indolent, to aggressive and rapidly progressive disease. Proliferation is a strong predictor for disease outcome. In routine clinical practice, Ki-67 expression is used as a measure of proliferation. However, several studies have documented a high degree of inter-laboratory and inter-observer variation with Ki-67 immunohistochemistry. Phosphorylation of histone H3 occurs specifically during mitosis and hence serves as a specific marker for cells in mitosis.Methods and resultsWe investigated phosphohistone H3 (PHH3) immunohistochemistry as a proliferation maker in 28 tissue biopsies of MCL and compared the PHH3 results (as evaluated by direct microscopic visualisation and image analysis-aided scoring) with morphological subtyping, mitotic counts and Ki-67 index. We found PHH3-mitotic count was about sixfold higher than H&E-mitotic count (mitoses in 10 high power fields). Furthermore, PHH3-mitotic count in aggressive morphological variants of MCL was significantly higher than in usual MCL. The PHH3-mitotic count showed a strong linear correlation with PHH3-mitotic index (percentage positive cells).ConclusionsWe found PHH3 immunohistochemistry, a reliable mitosis-specific marker, in MCL. Performing precise counts and evaluating precise proliferation indices is easier with PHH3 immunohistochemistry. This contrasts with the conventional estimation of Ki-67 percentages by ‘eye-balling’.