The potential for technology to aid quantitative assessment in cutaneous T cell lymphoma

Primary cutaneous T-cell lymphomas (CTCLs) consists of a heterogenous group of non-Hodgkin lymphomas defined by clonal proliferation of T cells in the skin, with Mycosis Fungoides (MF) and Sezary Syndrome (SS) most common, manifesting as patches, plaques, nodules, or erythroderma [1]. MF and SS are staged with the Tumor, Nodule, Metastasis, Blood (TNMB) model, which has prognostic significance but does not account for percent body surface area (BSA) involvement [2]. Modified Severity-Weighted Assessment Tool (mSWAT) scores are used in the clinical and research setting to assess skin tumor burden in MF and SS patients and are derived from manual estimates of percent BSA affected, weighted by lesion type: patch (x1), plaque (x2), or nodule (x4) [2]. mSWAT scores are calculated from a clinician’s visual assessment of affected BSA and lesion type and have considerable interrater variability, making direct comparisons challenging [2]. The emergence of new noninvasive imaging technologies, such as 3 D-total body photography (3 D-TBP), presents a unique opportunity to use high-resolution images to aid in clinical examination and objective monitoring of disease progress [3].

In this pilot study, we assessed the potential for using total body photography to improve standardized assessment of disease extent in cutaneous lymphoma, improving objective assessment of mSWAT scores.

This retrospective study was approved by the Memorial Sloan Kettering Cancer Center (MSK) Institutional Review Board. Ten patients from the cutaneous lymphoma clinic at MSK who had undergone 3 D-total body photography (3 D-TBP) and in-person mSWAT assessment on the same day from 08/2020–04/2021 were included in the study. 3 D-TBP was taken using Vectra WB360 whole body imaging system that captures entire exposed body surfaces in single capture with 92 cameras. Demographics, medical and treatment history, and previously documented mSWAT scores assessed from in-person clinic visits were collected from patient charts. For the ten patients, two medical students manually segmented patch, plaque, and tumor areas of cutaneous lymphoma on 3 D-TBP images using the program Vectra Analysis Module (VAM), and then mSWAT scores were derived from automated estimation of affected BSA by lesion type (Figure 1). Two dermatologists blinded to the original mSWAT scores approximated traditional mSWAT scores from visual assessment of the 3 D-TBP images. Consistency between mSWAT scores were analyzed using Pearson’s correlation coefficient. Agreement between mSWAT scores were analyzed using Lin’s concordance correlation coefficient (CCC), including reproducibility of in-person mSWAT scores using 3 D-TBP photos (by the dermatologist who rendered the in-person visit).

Figure 1. Sample photographs of a patient with cutaneous lymphoma in Vectra Analysis Module (VAM) software. (A) Patient with cutaneous lymphoma plaques on the back, and (B) manual demarcation of areas of plaque. Demarcated body surface area is automatically calculated in VAM and divided by the total body surface area to get the body surface area percentage affected. This process is repeated for patches and nodules, and the mSWAT score is manually calculated.

Ten patients with CTCL were included in the study. Demographics and CTCL details are provided in Supplementary Table. Median age (IQR) was 59 (15.8) years. Females represented the majority of the patients (n = 8, 80%). Of the ten patients, eight were white (80%), one black (10%), and one other (10%). All patients had a diagnosis of mycosis fungoides, with five patients (50%) with stage I disease, four patients (40%) with stage II, and one patient (10%) with stage IV. Based on in-person mSWAT scoring, one patient (10%) had patch stage only disease, and two patients (20%) had plaque stage only disease, six patients (60%) had patch and plaque stage disease, and two patients (20%) had patch, plaque, and tumor stage disease.

In-person mSWAT scoring was performed by two dermatologists (n = 3, n = 7). Overall, dermatologists’ in-person mSWAT assessment was highly similar to his or her visual assessment of mSWAT from 3 D-TBP images (Lin’s CCC = 0.95, 95% CI: 0.81, 0.99).

mSWAT scores between two medical students calculated by visual segmentation of 3 D-TBP images were highly correlated (Pearson’s correlation coefficient = 0.99, 95% CI: 0.90, 0.99) and in high agreement (Lin’s CCC = 0.94, 95% CI: 0.86, 0.98) (Table 1). mSWAT scores calculated by each of the two dermatologists from visual assessment of 3 D-TBP images were highly correlated (Pearson’s correlation coefficient = 0.92, 95% CI: 0.70, 0.98) and in high agreement (Lin’s CCC = 0.83, 95% CI: 0.54, 0.94). Pairwise comparison of mSWAT scores calculated from 3 D-TBP by the two dermatologists and two medical students showed higher correlation (Pearson’s correlation coefficient = 0.91, 0.83, 0.82, 0.74) but less agreement (Lin’s CCC = 0.44, 0.48, 0.68, 0.71).

Table 1. Pearson correlation and Lin’s CCC matrices of mSWAT scores calculated from 3 D-TBP, either visually by the dermatologists or manual segmentation by the medical students.

Dermatologist 1	Dermatologist 2	Medical student 1	Medical student 2
Pearson Correlation
1.00 (1.00, 1.00)	0.92 (0.70, 0.98)	0.83 (0.41, 0.96)	0.74 (0.21, 0.93)	Dermatologist 1
	1.00 (1.00, 1.00)	0.91 (0.64, 0.98)	0.82 (0.39, 0.96)	Dermatologist 2
		1.00 (1.00, 1.00)	0.98 (0.90, 0.99)	Medical student 1
			1.00 (1.00, 1.00)	Medical student 2
Lin’s CCC
1.00 (1.00, 1.00)	0.83 (0.54, 0.94)	0.44 (0.11, 0.68)	0.48 (0.07, 0.75)	Dermatologist 1
	1.00 (1.00, 1.00)	0.68 (0.35, 0.86)	0.71 (0.28, 0.90)	Dermatologist 2
		1.00 (1.00, 1.00)	0.94 (0.86, 0.98)	Medical student 1
			1.00 (1.00, 1.00)	Medical student 2

Parentheses contain the 95% confidence intervals.

The gold standard mSWAT calculation is currently through in-person assessment [4]. However, estimation of body surface area is highly variable, as evidenced by high interrater variability in mSWAT scores [2]. 3 D-TBP offers a higher resolution platform to visualize skin and may allow for standardized, objective assessment of mSWAT scores. In our study, dermatologist’s in person mSWAT and mSWAT visually assessed from 3 D-TBP were highly correlated and consistent, suggesting that this technology may be deployed in non-dermatology clinics and reviewed remotely by dermatologists for clinical trials (to reduce study visits) or other purposes. This pilot study demonstrates that mSWAT scores can be reliably calculated using 3 D-TBP and that automated segmentation may provide accurate estimates of disease in cutaneous lymphoma. 3 D-TBP can potentially be utilized in the clinical setting for long-term clinical monitoring of treatment response in patients with MF or SS.

A limitation of 3 D-TBP is that some body areas are unable to be adequately assessed—for example, 3 D-TBP does not capture the bottom of the feet, the buttocks or genital regions are unable to be assessed in patients wearing underwear, and in cases of patients with longer hair, the scalp is unable to be examined in detail. To address some of these limitations in our study, we reviewed patient charts and other clinical photographs to assess involvement of genitals/buttocks or bottom of the feet, but the optimal workflow for imaging and assessment would need to be defined through a prospective study of paired mSWATs performed in-person and using 3 D-TBP. In addition, in future studies, efforts should be taken to include a more diverse population, particularly patients with skin of color, in which areas of erythema or rash may be harder to visualize in photographs [5].

Calculation of mSWAT scores through in-person visual assessment may be prone to human error. Our study suggests that 3 D-TBP may be a promising tool in assessing standardized quantification of mSWAT scores. Use of 3 D-TBP in calculating mSWAT scores should be further evaluated in larger studies.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This study was supported by a grant from the National Cancer Institute/National Institutes of Health [P30-CA008748] made to the Memorial Sloan Kettering Cancer Center. Funding source had no oversight over the design and conduct of the study, collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

This study was supported by a grant from the National Cancer Institute/National Institutes of Health [P30-CA008748] made to the Memorial Sloan Kettering Cancer Center. Funding source had no oversight over the design and conduct of the study, collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.