Are MicroRNAs Key to Developing Biomarkers for Cutaneous T-Cell Lymphoma?

Abstract

MicroRNAs are noncoding molecules that induce epigenetic modulation, and they have been involved in tumorigenesis of solid and hematologic malignancies, including cutaneous T-cell lymphoma. MicroRNAs appear to play a role in cutaneous T-cell lymphoma pathogenesis and in disease progression. We comment on recent efforts to develop microRNA classifiers that improve cutaneous T-cell lymphoma diagnosis and predict disease course.

Cutaneous T-cell lymphomas (CTCLs) are rare skin cancers of skin-homing T cells. Among the different CTCL sub-types, mycosis fungoides (MF) is the most common, followed by CD30 lymphoproliferative disorders, Sézary syndrome, subcutaneous panniculitis-like T-cell lymphoma, and primary cutaneous γδ T-cell lymphomas. Despite recent progress in our understanding of this disease, there remains a pressing need for molecular biomarkers (Park et al., 2017).

Early stages of MF may resemble eczema or psoriasis clinically and histologically. There are few, if any, routinely used biomarkers that definitively distinguish CTCL from benign dermatoses. Diagnosis of CTCL often requires multiple skin biopsies over a median of 5 years. After diagnosis, there are few, if any, biomarkers that predict clinical course. Most MF patients have an indolent course, but at least 20% of MF patients progress to more advanced disease despite treatment. Predictive biomarkers could enable clinicians to offer more aggressive and effective therapies to the patients who might derive the most benefit.

MicroRNAs can be molecular drivers of cancer pathogenesis. MicroRNAs are small noncoding RNA molecules that induce epigenetic modulation by down-regulating gene expression. MicroRNAs inhibit translation or induce degradation of messenger RNAs. Depending on the genes involved, microRNAs may affect apoptosis or cell proliferation. Thus, microRNAs could function as oncogenes (oncomiRNA) when they are overexpressed or as tumor suppressor genes (tumor-suppressor microRNA) when they are underexpressed. The same microRNA has been reported to act as a tumor-suppressor microRNA and an oncomiRNA depending on the environmental and cellular context (Papadavid et al., 2016). Because microRNA expression profiles (micro-RNA signatures) can be cancer subtype-specific, these signatures may serve as putative biomarkers for solid malignancies and leukemia/lymphomas, including CTCL.

MicroRNA profiling of CTCL has been done in both skin and blood. Multiple groups have identified putative onco-miRNAs that include miR-155, miR-21, miR-199, miR-214, and miR-486, and miR-22 is a putative tumor-suppressor microRNA (Garaicoa et al., 2016). The molecular basis of dysregulated microRNA expression remains unclear, but it may relate to dysregulated signaling in CTCL, including that mediated via the JAK/STAT pathway (Sibbesen et al., 2015). Multiple groups have suggested that microRNA signatures can be used to distinguish healthy skin or benign skin disease from CTCL, to differentiate CTCL subtypes, and to monitor treatment responses (Tables 1 and 2). No single microRNA appears to have sufficient specificity and sensitivity to serve as a biomarker. Investigators have increasingly turned to sets of three to five microRNAs, termed microRNA classifiers, as novel diagnostic or prognostic tools. Table 1 summarizes the most remarkable microRNA classifiers developed in CTCL, mainly in MF and/or Sézary syndrome, to date (Dusilkova et al., 2017; Lindahl et al., 2018; Marstrand et al., 2014; Ralfkiaer et al., 2011; Shen et al., 2018).

Table 1.

MicroRNA classifiers reported for CTCL diagnosis

Purpose	Reference	Platform Used	MicroRNAs Included in MicroRNA Classifiers		Number of Samples Tested	Sensitivity, %	Specificity, %
			Up-regulated	Down-regulated
Diagnosis	Ralfkiaer et al., 2011	miRCury LNA array (Exiqon, Vedbaek, Denmark), qRT-PCR based	miR-155	miR-203, miR-205	43 CTCL 60 benign	91	97
	Marstrand et al., 2014	TaqMan miRNA assays (Thermo Fisher Scientific, Waltham, MA), qRT-PCR based	miR-155	miR-203, miR-205	30 CTCL 48 benign	90	97.9
	Dusilkova et al., 2017	qRT-PCR based	miR-155	miR-203, miR-205, miR-22, miR-223	36 CTCL 11 benign	94	100
	Shen et al., 2018	Microarrays (Agilent, Agilent Technologies, Santa Clara, CA; and Affymetrix, Thermo Fisher Scientific, Waltham, MA), qRT-PCR based	miR-130b, miR-142-3p, miR-155	miR-200b, miR-203	158 CTCL 70 benign	Training: 98.3 Testing: 96	Training: 96 Testing: 72

Abbreviations: CTCL, cutaneous T-cell lymphoma; miR, microRNA; qRT-PCR, quantitative real-time reverse transcriptaseePCR.

Table 2.

MicroRNA classifiers developed for prognosis and disease monitoring

Purpose	Reference	Platform Used	MicroRNAs Included in MicroRNA Classifiers		Number of Samples Tested	Results
			Up-regulated	Down-regulated
Prognosis	Lindahl et al., 2018	qRT-PCR based	miR-106b-5p, miR-148a-3p, miR-338-3p	—	154 CTCL	MicroRNA classifier risk score: hazard ratio (95% of coefficient interval) = 13.5 (5.6e32.7)
	Shen et al., 2018	Microarrays (Agilent, Agilent Technologies, Santa Clara, CA; and Affymetrix, Thermo Fisher Scientific, Waltham, MA), qRT-PCR based	miR-155	miR-200b	108 CTCL	Sensitivity = 68.9% Specificity = 71.6% Addition of percentage ki-67 in skin biopsy sample to generate a new model: Sensitivity = 77% Specificity = 81.1%
Disease monitoring	Dusilkova et al., 2017	qRT-PCR based	miR-155	miR-203, miR-205	10 CTCL 11 benign	Increase/decrease of three-microRNA score predicted response

Abbreviations: CTCL, cutaneous T-cell lymphoma; miRNA, microRNA; qRT-PCR, quantitative real-time reverse transcriptaseePCR.

Shen et al. (2018) present a novel microRNA signature for diagnosis and prognostication of CTCL patients. This tool may improve patient care. Although this study has multiple strengths, the most important is the relatively large sample size: 158 cases of CTCL of various stages and 70 cases of benign inflammatory skin disorders that were separated into disovery and validation sets. Shen’s group performed an initial screening of micro-RNAs in a discovery phase. A total of 50 skin biopsy samples of CTCL and 20 biopsy samples of benign skin conditions were tested using two different microarray platforms. Candidate microRNAs were those whose expression changed more than 2-fold in either direction with P less than 0.05. Nine candidates were identified using both platforms (miR-27, miR-130b, miR-150, miR-155, miR-200b, miR-200b, miR-203, and miR-342-3p). An additional microRNA, miR-142-3p, was also included as a candidate because it was the most differentially expressed miRNA using the Agilent microarray platform (Agilent Technologies, Santa Clara, CA). These initial 10 candidate microRNAs were further assessed by reverse transcriptase-PCR in an independent cohort of 83 samples in the training phase. Among the 10 candidate microRNAs, 5 were differentially expressed using reverse transcriptase-PCR with P less than 0.05 in multilogistic regression analysis comparing CTCL and benign skin conditions. These five microRNAs, including miR-200b, miR-203, miR130b, miR-142-3p, and miR-155, were tested as the microRNA classifier. CTCL was diagnosed with a sensitivity of 98.3% and specificity of 96% in the training set. These results were validated with an independent data set with similar sensitivity (96%) and specificity (72%). A microRNA signature tested to distinguish CTCL subtypes fared less well. Comparing early stage MF and advanced stage and other CTCL subtypes, only 64% of the samples were classified correctly. Although none of the advanced MF or other CTCL samples were classified as benign skin conditions, 55% of the advanced MF cases were classified as other subtypes of CTCL.

MicroRNA expression profiles were also shown to have prognostic implications. Two microRNAs correlated with follow-up data from the training and validation sets, with a median duration of 90 months. These included miR-155 and miR-200b, which showed opposite directions of correlation. High expression of miR-155 and low expression of miR-200b correlated with worse prognosis. The prediction of 5-year overall survival was modestly improved with the addition of Ki-67 as an additional biomarker (Shen et al., 2018).

The microRNA signatures identified by Shen et al. (2018) include some microRNAs that have been functionally validated in CTCL or other cancers. Among all microRNAs, miR-155 is the most comprehensively investigated in CTCL, and there are ongoing clinical trials with miR-155 inhibitors in CTCL patients (NCT02603224). As mentioned before, STAT5 induces the expression of miR-155, which promotes proliferation of malignant T cells. Overexpression of miR-155 has been observed in skin biopsy samples of tumor-stage MF and in circulating malignant Sézary cells, but elevated miR-155 expression in early MF lesions has been variably detected. The duration of expression may be more important than expression level because prolonged overexpression of miR-155 can lead to uncontrolled proliferation, inhibition of apoptosis, and malignant transformation bridging inflammation and carcinogenesis (Garaicoa et al., 2016). This could explain why some studies did not find significant differences in miR-155 expression in benign skin conditions and early MF.

Down-regulation of different components of the miR-200 cluster has been associated with worse prognoses in many different malignancies. Shen et al. (2018) determined that miR-200b was a biomarker with prognostic significance in CTCL, as has been re-ported in epithelial malignancies. Down-regulation of miR-200 results in epithelial-mesenchymal transition and enhances tumor metastasis (Shen et al., 2018). The authors hypothesized that, through this mechanism, malignant lymphocytes escape from the epidermis in advanced stages of the disease. Alternatively, down-regulation of miR-200s (such as miR-200c) has been shown to increase expression of Jagged-1 protein, a cell surface ligand of the Notch pathway (Gallardo et al., 2015).

The signature identified by Shen et al. (2018) may help clinicians diagnose CTCL patients more quickly and personalize treatments for individual patients. However, as yet, there is no consensus microRNA signature that is in routine clinical use. One or two microRNAs may be shared in different classifiers but, in general, there is little to no overlap between microRNA signatures that have been described by different investigators. For example, a recent report from Lindahl et al. (2018) described a three-microRNA classifier that identified patients with early-stage MF who were at increased risk for disease progression. Using a cohort of 154 Danish patients, the microRNA classifier discriminated a high-risk cohort from a low-risk cohort with respect to disease progression. Progression-free survival and overall survival showed hazard ratios of 4.60 (95% confidence interval = 1.75−12.0) and 2.39 (95% confidence interval = 1.46−3.92), respectively (Lindahl et al., 2018). This microRNA classifier includes miR-106–5p, miR-148-3p, and miR-338-3p and has no overlap with the microRNAs identified in the Shen et al. study.

It is not clear why microRNA classifiers that have been identified are so heterogeneous. Different studies feature differences in study design, methodology used to query microRNA signatures (e.g., platforms arrays versus real-time PCR vs. next-generation sequencing), and statistical methods. We also cannot exclude the effect of other potential confounders. These include (i) the heterogeneity of CTCL; (ii) the use of different types of control samples (e.g., healthy skin, atopic dermatitis, eczema, psoriasis, discoid lupus); (iii) differences due to differences in patient ethnicities; (iv) different environmental factors; and (v) nonuniform sample sizes.

In conclusion, in light of novel microRNA-based therapies and for other reasons, there is a critical need to identify microRNAs that are clinically useful biomarkers or therapeutic targets in CTCL. MicroRNA classifiers such as those proposed by Shen et al. (2018) may improve patient care by aiding early disease diagnosis and linking patients to appropriately aggressive therapies. The collection of large cohorts of CTCL, a rare disease, for these types of molecular studies is incredibly valu-able. However, translating these findings to the clinic will require rigorous testing and, perhaps, a prospective multicenter study with a uniform method to measure microRNA expression.

Clinical Implications.

• MicroRNA classifiers may be useful for CTCL diagnosis, determining prognosis, and monitoring responses to therapy.

• miR-130b, miR-142-3p, miR-155, miR-200b, and miR-203 may help distinguish CTCL from inflammatory conditions.

• Dysregulation of miR-155 and miR-200b may predict decreased overall survival.