Chalkley Counting in Oral Tongue Squamous Cell Carcinoma : Does It have A Prognostic Value ?

Chalkley counting has been regarded as a relatively reliable method of quantifying tumor angiogenesis. In this study we investigated the reliability of Chalkley counting in quantifying tumor angiogenesis in oral tongue squamous cell carcinoma (OTSCC) using CD34; and tumor vasculogenesis using angiotensin converting enzyme, angiotensin II receptor 1 and angiotensin II receptor 2, in 32 OTSCC samples. Chalkley counting was performed by two independent observers. The averages of three ‘hot spot’ counts were compared with known prognostic factors. All four markers showed no correlation with any of the prognostic factors. When comparing the results from the two independent observers, the only marker shown to have a significant moderate correlation was CD34. The other three markers showed no significant correlation. The lack of statistical significance between the independent observers, and known prognostic factors with the four markers used, shows that Chalkley counting is not a reliable prognostic tool in OTSCC. DOI : 10.14302/issn.2576-6694.jbbs-19-2625 Corresponding author: Swee T Tan, ONZM MBBS PhD FRACS, Gillies McIndoe Research Institute, PO Box 7184, Newtown 6242, Wellington, New Zealand, Ph: +64 4 2820366, Email: swee.tan@gmri.org.nz Running title: Chalkley counting for oral tongue squamous cell carcinoma


Introduction
Oral cavity squamous cell carcinoma (OCSCC) is the 15th most common cancer worldwide [1] with vast geographical differences and greater incidence in developing countries [2,3].OCSCC affects males most commonly, in their fifth and sixth decades of life, although the incidence is increasing in women and those under the age of 45 [3].Risk factors for OCSCC include alcohol abuse, tobacco smoking and betel quid chewing [3].
The prognosis of OCSCC depends on tumor stage and other factors [2,4,5] including the extent of tumor angiogenesis -the development of new vessels from pre-existing blood vessels [6,7].The observation that tumor growth and metastasis are dependent on tumor angiogenesis led to its quantitation to determine tumor-related prognosis, with studies confirming this association [8,9].
Chalkley counting has been regarded as a relatively reliable method of quantifying tumor angiogenesis [9][10][11].This standardized method counts immunohistochemically stained endothelial cells within the tumor by using a 25-point Chalkley graticule and orientating it to overlap the highest number of stained microvessels [9][10].Quantifying microvascular density by selecting areas with the most stained vessels -the neovascular 'hotspots', then counting distinct microvessels within a microscopic field of view [10].
Others [9,12] find quantifying angiogenesis in breast cancer by Chalkley counting of CD34+ endothelial cells as an independent prognostic factor.Waengertener et al. [11] also demonstrated an association between microvascular density quantified by Chalkley counting and the survival of patients with gastrointestinal stromal tumors.However, there remains no consensus on the accuracy and usefulness of Chalkley counting as a method of quantifying tumor angiogenesis, and hence prognosis.
Although Chalkley counting is a reasonably simple and commonly used histopathology procedure [13], its dependence on relatively subjective and observer-dependent selection of the microvascular 'hotspots' has led to its reliability in quantifying angiogenesis in breast cancer being questioned by the College of American Pathologists [14].
Physiologically, the renin-angiotensin system (RAS) regulates blood pressure and involves the conversion of angiotensinogen to angiotensin I (ATI) by renin.ATI is then converted by the angiotensin converting enzyme (ACE, also known as CD143), to angiotensin II (ATII) which acts on angiotensin II receptor I (ATIIR1) and angiotensin II receptor 2 (ATIIR2) [15].ACE is a marker for embryonic stem cell-derived hemangioblast differentiation [16].It regulates hemangioblast expansion and differentiation into either hematopoietic cells or endothelial cells via activation of ATIIR1 and ATIIR2, respectively.
Tumor vascular mimicry is a de novo blood vessel formation from cancer stem cells, rather than pre-existing endothelial cells [17].Positive human control tissues used to confirm the specificity of the primary antibodies were liver for ACE and ATIIR1, and kidney for ATIIR2 (data not shown).

Quantitation of Tumor Angiogenesis and Tumor Vasculogenesis by Chalkley Counting
The The graticule was then orientated so that as many points as possible were on or within the positively stained microvessels.The average counts of both observers' three 'hotspots' gave the total vascularity score used in the analyses.

Statistical Analysis
To determine statistical significance a Pearson's correlation was performed using IBM SPSS v22.

CD34 Chalkley Counting
When comparing the CD34 Chalkley counts between the two independent observers (P1 and P2), we could only detect a moderate correlation across the counts (Pearson's r = 0.498, p < 0.001).Pearson's correlation showed no significant correlations between the CD34 Chalkley counts and the prognostic factors of OTSCC chosen for this study (Table 1).

ACE Chalkley Counting
When comparing the ACE Chalkley counts from the two independent observers (P1 and P2), no significant correlation was detected (Pearson's r = 0.222, p > 0.05).Pearson's correlation showed no significant correlations between the ACE Chalkley counts and the prognostic factors of OTSCC chosen for this study (Table 2).
Pearson's correlation showed no significant correlations between ATIIR1 Chalkley counts and the prognostic factors of OTSCC chosen for this study (Table 3).Pearson's correlation showed no significant correlations between the ATIIR2 Chalkley counts with the prognostic factors of OTSCC chosen for this study (Table 4).

Discussion
Chalkley counting as a method of quantifying tumor angiogenesis and tumor vasculogenesis to prognosticate cancer has been widely used across different cancer types, including breast [9,13,19] and gastrointestinal [11] cancers.Hansen et al. [13] investigated different methods of quantifying tumor angiogenesis in breast cancer and reported that Chalkley counting produces the least observer variability.They used this method to study 836 breast cancer patients and concluded that Chalkley count is a reliable and independent prognostic tool for breast cancer [9].
Despite this earlier study indicating Chalkley count having the least observer variability in selecting the microvascular 'hotspot' (the most observer-dependent step), the College of American Pathologists regards quantifying microvessel density by Chalkley counting as being an unreliable prognostic tool for breast cancer [14].
Vascular mimicry leads to greater perfusion in cancer leading to tumor growth and metastasis [17].We had therefore chosen ACE in this study on OTSCC to quantify tumor vasculogenesis.ATIIR1 and ATIIR2 were also selected given their involvement in putative stem cell differentiation [16].
The  [19][20].Furthermore, the appropriateness of using ACE, ATIIR1 and ATIIR2 as markers for tumor vasculogenesis, at least in OTSCC, remains to be conclusively determined.These factors contribute to a decrease in the overall reliability and validity of this method and bring in to question its role in the prognostic setting.
We applied Chalkley counting to quantify tumor angiogenesis using CD34 and tumor vasculogenesis using ACE, ATIIR1 and ATIIR2 but found this to be unreliable in OTSCC.Its weakness relates to the reproducibility (also known as reliability) of its results.
Reliability is a criterion that is applied universally to measuring instruments and is generally obtained by correlating the results of repeated measurements on the same things by both the same and different people (co-efficients of equivalence) and/or at similar and different times (coefficients of stability).When both times and observers are different it yields the coefficient of stability and equivalence.As well as being important for obvious reasons, reliability is critical for its relationship to validity (the degree to which the measurement actually reflects the characteristics of what it is measuring).In almost all cases reliability indicates the maximum possible level of validity that can be obtained.Acceptable levels of reliability begin to be reached when its correlations are 0.7 or more -meaning more than about 50% of the variance is accounted for.
Depending on the task, however, levels of reliability may need correlations in the range of 0.9 (meaning 80%+ of the variance is required to be accounted for).Averaging observations from a number of observers or a number of repeats will increase the reliability estimates to levels which can be predicted from the application of the Spearman-Brown Formula.In our observations we recorded correlations no greater than 0.5.These results indicate that Chalkley counting is not suitable for our measurements procedures.
tumor was marked on the slide by an anatomical pathologist (HDB) and tumor angiogenesis was quantified by two independent observers (P1 and P2) by counting the CD34+ endothelial cells on the microvessels within and immediately adjacent to the tumor.The extent of tumor vasculogenesis was similarly quantified by counting cells lining the microvessels that stained positively for either ACE, ATIIR1 or ATIIR2.Each slide was read by two observers at 400x magnification to subjectively select three 'hotspot' areas within each tumor that showed the greatest number of distinct positively stained microvessels.Each 'hotspot' was then assessed using a 25-point Chalkley graticule (Olympus, Tokyo, Japan), in a 200x field on a light microscope (10x eyepiece, 20x objective, area 0.196 mm 2 ) (Olympus).
staining (Fig.1A) confirmed the presence of SCC on the slides.To determine if Chalkley counting was a reliable method of quantifying tumor angiogenesis and tumor vasculogenesis, Chalkley counting on CD34 (Fig.1B, brown), ACE (Fig.1C, brown), ATIIR1 (Fig.1D, brown) and ATIIR2 (Fig.1E, brown) by two independent observers (P1 and P2) was performed.The averages of their three 'hot spot' counts were compared with known prognostic factors chosen for this study: tumor TNM stage, and clinical stage (Suppl.