The study was approved by the Ethical Review Committee of Fourth Medical Center of PLA General Hospital. Written informed consent was signed after the patients were being informed of the research content and research methods.

By fully searching the clinical database of our admitted patients, a total of three cases were included in the study that met the diagnostic criteria of TLC with pathological confirmation. Clinical data and photographs were obtained and desensitized for the protection of patients’ privacy. All the three patients had previously undergone tissue biopsy for the differential diagnosis of the lesions. The biopsy tissues were fixed in formaldehyde, embedded in paraffin, and preserved in the Department of Pathology of Fourth Medical Center of PLA General Hospital. With the patients’ consent, the specimens were obtained and sliced for further immunohistochemical staining.

For the obtained specimens, sections were incubated with DNase-free proteinase K (20 μg/ml; P1120, Solarbio, China) at 37°C for 30 min for antigen retrieval. Then, sections were incubated with 0.1% Triton X-100 (HFH10, Invitrogen, USA) in phosphate buffered saline (PBS) for 30 min to penetrate cell membrane. After blocking with 5% goat serum (16210064, Gibco, USA) in PBS for 30 min, the sections were incubated with corresponding antibodies overnight at 4°C. After incubation, sections were rinsed three times in PBS for 5 min each and then incubated with corresponding goat anti-rabbit or goat anti-mouse secondary antibodies conjugated with horseradish peroxidase for 1 h at room temperature. Sections were washed with PBS and then exposed to 3,3′-diaminobenzidine (DAB) staining (DA1016, Solarbio, China) following the manufacturer’s instructions. Finally, the slices were observed and photographed using a panoramic scanner from 3DHISTECH CaseViewer.

A comprehensive literature review was conducted by searching the publications in PubMed (http://www.ncbi.nlm.nih.gov) and Embase (https://www.embase.com/) published between 1977 and 2022. The search terms used were “trichilemmal carcinoma”, “tricholemmal carcinoma”, and “tricholemmocarcinoma”. References not indexed in PubMed or Embase were also traced for a complete record. Specifically, the Cochrane Library (https://www.cochranelibrary.com/) and Prospero (https://www.crd.york.ac.uk/prospero/) were explored to obtain the TLC-relevant meta-analyses. The databases of ClinicalTrials.gov (https://clinicaltrials.gov/) and National Cancer Institute (https://www.cancer.gov/) were searched for any clinical trials on TLC. Translation of articles in languages other than English was executed via Google Translate (http://translate.google.com). The cutoff date was 7 July 2022. Eighty-six publications worldwide were identified from PubMed and Embase containing a total of 228 cases. No TLC-relevant meta-analyses or clinical trials were identified.

Excel and SPSS were utilized for processing information. Missing values were ignored. Descriptive statistics were utilized. For categorical variables, frequencies and the percentage were computed, and Chi-square test or two-sided Fisher’s exact tests were performed. Continuous variables were first tested for normality by the Kolmogorov–Smirnov test and then for homogeneity by the variance test. Normally distributed data were presented as mean ± standard deviation and tested using t test. Non-normally distributed data were presented as the median and interquartile range and compared using the Mann–Whitney U-test. Covariates with a P-value < 0.05 in the univariate analysis were chosen for the multivariate analysis, where logistic regression was used to determine each variable with the recurrence of TLC. The 0.05 level of confidence was accepted as a significant difference.

All the reported cases were collected to analyze the characteristics of TLC. In total, 231 cases (including three cases in this article) were included. Because of the infrequent reporting or tracking of TLC occurrences, the actual incidence could not be accurately determined. The cumulative number of cases has grown slowly since the mid-1990s but been increasing rapidly since 2014 with a gradually upward growth curve ( Figure 4A ). The year 2020 (45, 19.48%), 2018 (37, 16.02%), and 1994 (28, 12.12%) were the top three years with the highest volume of published TLC cases ( Figure 4A ). All continents except Antarctica have cases, with most of the cases located in Asian (112, 48.48%) ( Figure 4B ). China ranked first (50, 21.65%), followed by the USA (41, 17.75%), UK (40, 17.32%), Japan (30, 12.99%), and Korea (12, 5.19%).

The clinical demographic information of the 231 cases is summarized in Table 1 . A total of 121 cases were men, slightly more than 110 of women ( Table 1 ). The age ranged from 9 to 95 years, and there was no statistical difference between genders (t = 1.623, P = 0.1064). However, when grouped by every 20 years, the age distribution between genders was inconsistent (c² = 11.255, P = 0.016). The age group with the highest proportion of men was 61–80 with 51 cases (53.68%), higher than that of the women (24 cases, 32.43%, P < 0.05). Meanwhile, the proportion of female patients over 80 years old was the highest (31 cases, 41.89%), which is statistically different from that of the male patients (19 cases, 20.00%, P < 0.05).

Most common location was the head and neck (194 cases, 83.98%). The remaining cases were distributed on leg, chest, hand, arm, abdomen, shoulder, armpit, buttocks and perineum, and back. There was no statistical difference in the distribution between genders (c² = 8.288, P = 0.508).

The phenotype of TLC was not always typical. In addition, TLC may be misdiagnosed or overlooked due to the co-occurrence of other diseases. Three cases presented with actinic keratosis simultaneously, two cases on ear, and one case on temple (4–6). The lesion on the abdomen of a 76-year-old woman was diagnosed as TLC arising in seborrheic keratosis (7). A case of TLC in facial sebaceous glands was accompanied by Syringocystadenoma papilliferum (8). There were also cases of TLC with primary cutaneous anaplastic large-cell lymphoma, a ductal eccrine component, nevus sebaceous, or a proliferating trichilemmal cyst (9–12). Some cases were combined with other carcinoma at the lesions, four cases with BCC (13, 14), two cases with Bowen’s disease (15, 16), and one case with Merkel cell carcinoma (17).

Manifestations of TLC ranged in diameter from 0.3 to 37 cm, and the thickness ranged from 0.30 to 10.00 cm. Diameter was not statistically different between genders (Z = 0.442, P = 0.658). On the contrary, male cases appear to have superiority in thickness (median, 2.50 cm) than that of female cases (median, 0.80 cm, Z = 2.106, P = 0.035). The duration had a median time of 13.0 months; 12.0 months in men and 24.0 months in women but with no statistical difference (Z = 1.525, P = 0.127).

Furthermore, we explored factors that may be related to the size or duration of TLC ( Figure 5 ). The diameter of TLC had an obvious linear relationship with its thickness (r = 0.7887, P < 0.001) and a relatively weak linear relationship with the duration (r = 0.3325, P = 0.0033) but had no linear relationship with the age (r = 0.0565, P = 0.4999). In addition, no linear relationship was found between thickness with age (r = 0.2418, P = 0.2341), thickness with duration (r = 0.1986, P = 0.4294), or duration with age (r = 0.1905, P = 0.0807).

Another thing to note is that nine cases were accompanied by regional lymph node metastasis, with five men and four women.

The pathogenesis of TLC has not been clearly elucidated. A total of 87.01% of TLCs (201 of the 231 cases) occurred in the sun-exposed sites (head, neck, arm, and hand), 6.7 times of the non-exposed sites, suggesting the significant role of UV in the pathogenesis of TLC. Ten cases (4.33%) had previous carcinoma: four cases with breast cancer (including Case 2) (18–20), five cases with BCC (two of whom also had SCC) (13,) (21–24), and one case with both SCC and Bowen’s disease (25). There were four and six cases in men and women, respectively (c² = 0.228, P = 0.633). Moreover, locally present skin lesions, trauma, or scarring may also be associated with the progression of TLC. Two cases were secondary to a previous site of actinic keratosis in neck and forehead, respectively (13, 25). TLC was also secondary to a previous proliferating trichilemmal (26), psoriasis (21), or cutaneous horn of the location (27). Local inappropriate physical or chemical stimulation may trigger the occurrence of TLC. A 36-year-old woman developed TLC after receiving laser treatment in the postauricular area (28). Prolonged topical use of hair removal cream on armpits also triggered TLC, with metastases to regional lymph nodes (29). In two cases, TLC was secondary to scars caused by burns 5 years or even 42 years ago (30). Ionizing radiation could induce carcinogenesis by inducing DNA damage and gene mutation. A 67-year-old man, who had been exposed to multiple x-rays (50–60 times in total) and CT of the chest, developed a 1.5-cm TLC with regional recurrence and distant metastasis after resection (31). However, it should be noted that the role of these factors in the occurrence and development of TLC may only be hypothesized now and remains to be further determined, given that the number of related cases was all small, which could very well be a random event.

Studies have shown that advanced age, Caucasian, male sex, and a chest organ transplant are the risk factors for skin carcinoma after solid organ transplantation (32). Here, a total of four cases from 39 to 63 years underwent transplantation [two kidney transplants (33, 34), one heart transplant (35), and one unknown (36)], all were men with three Caucasians and one Asian. Because immunosuppression may affect the phenotype of TLC, the true incidence of TLC secondary to solid organ transplantation may be higher (37).

Additional risk factors of TLC also include genetic disorders. As an autosomal recessive genetic skin disorder, patients with xeroderma pigmentosum (XP) have impaired DNA repair ability (38, 39). A 25-year-old man with XP presented with TLC in cheek and eyelid (40). Another 9-year-old girl with XP also suffered from TLC in the nose (41). Both cases were much younger than the average age, suggesting XP as a strong risk factor for TLC. The impaired DNA repair ability of XP was considered to be related to the changes of p53 to some extent (42). As a key anti-cancer protein with growth inhibitory function, p53 can regulate the cell cycle and promote cell apoptosis or senescence, thus inhibiting tumorigenesis (43). Studies have found that the total loss of chromosome arm 17p (where the TP53 gene resides) was found in a TLC case, leading to the blocked expression of p53, which may be associated with the marked increase of the proliferation fraction and thus contribute to the development of TLC (12).

With the same tissue origin as trichilemmoma, TLC has similar histologic features but varies by atypical nuclei and high mitotic index (44). Centered on a pilosebaceous unit, TLC generally presents a typical histopathology of trichilemmal keratinization, a peripheral palisading pattern and focal necrosis (41, 45). The growth of TLC always presents lobular and invasive, characterized by classical large, polygonal, transparent tumor cells with eccentric nuclei but without hair follicle differentiation (41, 46). In some cases, TLC was found attached to the epidermis and infiltrating the hair follicle within the dermis, with the infiltrative lobules of clear cells where trichilemmal keratinization can be found (47).

In general, the pathological diagnostic criteria of TLC are not uniform and varied among different authors and doctors, but the following criteria for diagnosing are accepted and recognized by the majority: (1) PAS-positive glycogen within neoplastic cells, (2) folliculocentricity, (3) peripheral palisading of clear cells, (4) a prominent Periodic Acid Schiff Diastase (D-PAS)–positive basement membrane, (5) trichilemmal keratinization, (6) lobular architecture, and (7) the presence of pre-existing trichilemmoma (30, 48). TLC may be confused with other skin cancer, such as SCC with clear cell differentiation, BCC with keratin cysts and peripheral palisading cells within the basaloid islands, or even the keratoacanthoma, which was benign and could resolve spontaneously (27, 49–51). Given the similarity of cell origin and pathological phenotype, the accurate diagnosis of TLC from other tumors originated from the skin and the adnexa may sometimes be difficult and relies further on stains other than HE staining.

In general, stains were not performed unless the differential diagnosis from other mimickers was necessary. From the statistics of all published cases, PAS staining was used in 73 cases, all of which were positive. Mucicarmine staining was applied in 23 cases, and all samples were negative. All the other 53 pathological indicators were counted and sorted into eight categories, with an average frequency of 7.06 ( Figure 6 ). Using the frequency ≥ 7 as filter criterion, there were 17 remained indicators. Furthermore, the top six indicators of positive and negative rates, respectively, are listed in Figure 7 . It is worth noting that Pan CK ranked first with a frequency of 35 times, but its differential diagnostic value for skin tumors with clear cell changes was limited, because Pan CK was always positive in the mimickers, except for tumors arising from different lines of differentiation, such as melanoma, which, in turn, may not be difficult to diagnose due to its distinct appearances in HE staining (52).

As a glycoprotein localized on the surface of cell membrane, CD34 is involved in the adhesion of cell–cell and cell–extracellular matrix (53). In addition to being used as a marker of hematopoietic stem cells, CD34 is also a clue to the differentiation from the outer root sheath and follicular, which is of great significance for the diagnosis of TLC (54–56). Here, a total of 12 cases of TLC were stained for CD34, of which 10 cases (83.33%) were positive. The staining of CEA in TLC was usually negative, and, of the 29 cases, only two cases were positive ( Figure 7 ). Moreover, EMA was widely adopted with a frequency up to 32 times, with half positive and half negative, suggesting that EMA may not be a suitable marker for identifying TLC ( Figure 6 ).

A total of 212 cases were treated by surgical excision, among which 16 cases (7.55%) experienced a recurrence. Mohs micrographic surgery (MMS) has been widely used in the resection of carcinoma especially in the head and face (57, 58). Sixteen cases of TLC were resected by MMS, of which two cases had tumor recurrence, namely, 77-year-old man and 65-year-old man both in cheek (23, 59). The recurrence rates of surgical excision and MMS were not statistically different (c² = 0.052, P = 0.820). Another recurrence occurred in a 64-year-old man who refused treatment (60). Excision margins ranged from 1 to 20 mm, with no linear relationship with TLC diameter or thickness (r = 0.2852 and 0.3395, P = 0.1982 and 0.4028, Figure 8 ). As for MMS, there was only one documented resection margin, which was 8 mm.

Other interventions have also been reported. Imiquimod cream (5%) successfully healed a 90-year-old woman with a lesion of 2.5 × 2.0 × 0.39 cm on the cheek (61). Chemotherapy was used to treat a patient with TLC with primary cutaneous anaplastic large-cell lymphoma (9). Three cases underwent additional radiotherapy after excision (62, 63). Two cases received both chemotherapy and radiotherapy after excision. A 27-year-old man successfully relieved his 5-cm lesion on the neck (26). However, the lesion in 64-year-old man’s leg still recurred after treatment (58).

Nineteen cases were recorded with recurrence of TLC. We further analyzed the risk factors for TLC recurrence with cases divided into recurrence or non-recurrence groups ( Table 2 ). There was no statistically difference in sex, age, location, thickness, regional lymph node metastasis, or regional non-carcinoma lesions (P > 0.05). Diameters and duration were much higher in the recurrence group (Z = 2.711 and 2.073, P = 0.007 and 0.038), with carcinoma history more frequently occurred (c² = 4.992, P = 0.025). Further logistic regression analysis revealed that carcinoma history was the independent risk factors for the recurrence of TLC, with odds ratio (OR) of 5.448 and 95% CI of 1.428–20.778 (P = 0.013, Table 3 ).