Abstract
Background/Aim: Squamous cell carcinoma (SCC) is highly prevalent in kidney transplant patients (KT). It is characterized by the presence of an inflammatory infiltrate. In this study, we examined the presence of similar infiltrates in intact skin, which could be regarded as a precancerous step. Patients and Methods: We retrospectively analyzed skin biopsies of 19 non-transplanted patients with a diagnosis of SCC or basal cell carcinoma (BCC) and 17 KT with either SCC or BCC. Results: KT showed increased inflammatory infiltrate in the subepithelial region, compared to non-transplanted patients. The density of basal cell nuclei was also different among the four groups with an interaction effect between tumor type and transplantation. The extent of inflammatory infiltrates did not correlate with the eGFR and proteinuria. Conclusion: KT with a non-melanoma skin cancer show increased intact skin inflammatory infiltrate and alterations in the density of the basal cell layer compared to non-transplanted patients.
The role of inflammation in cancer development and progression is complicated. In general, inflammation encompasses a series of immune cell types and pathways which respond to infectious agents and tissue damage by mobilizing both the innate and adaptive arms of the immune system. However, accumulating evidence from more than a decade's worth of studies has revealed that inflammation has opposing actions regarding cancer: pro-tumorigenic and anti-tumorigenic (1).
Depending on the initiating event (i.e. microbial, tissue injury), acute inflammation is characterized by activation of localized myeloid cells (i.e. macrophages, dendritic cells, etc.) in tandem with TH1- and TH17-helper T cells (2, 3). These cells collectively act to produce cytokines and inflammatory mediators. Myeloid cells produce and secrete cytokines and mediators such as tumor necrosis factor alpha (TNF), interleukin-2 (IL-2), interleukin-6 (IL-6), and interleukin-23 (IL-23) (3). Lymphoid derived T helper cells produce interferon-γ (IFN-γ), interleukin-17 (IL-17), interleukin-21 (IL-21), and interleukin-22 (IL-22) (3). The role of acute inflammation appears to be less important for tumor development, but it may create a self-propagating inflammatory environment in the presence of cancerous cells, although many details remain uncertain (4).
Sustained, chronic inflammation, driven by tumor associated “stress” (i.e. ROS, tumor metabolism, cell death, corrupted microbial danger signals), is thought to perpetuate a permissive environment for tumor growth (1). Chronic inflammation involves many of the same cell types and soluble mediators also involved in acute inflammation, as well as regulatory T cells, Th2 cells, and pro-tumorigenic cytokines and inflammatory mediators such as IL-4, IL-6, IL-10, IL-13, and transforming growth factor beta (TGF- β) (4, 5).
Some of the cytokines, inflammatory mediators, and cells are pro-tumorigenic (i.e. TNF- α, IL-1, IL-6, IL-11, IL-17, IL-18, IL-21, IL-22, and IL-23) whereas others are anti-tumorigenic (i.e. IFN-γ and IL-12) (4). TGF- β is an exception as it can be both anti-and pro-tumorigenic by inhibiting keratinocyte proliferation and tumor inflammation while favoring metastasis (4). Most of these interleukins influence tumor growth and survival genes by acting on STAT3 and NF- ĸB transcription factors in tumor cells (4). The cellular repertoire of the tumor microenvironment also influences tumorigenesis. In general, exclusion or inactivation of cytotoxic T cells in the tumor microenvironment also favors tumorigenesis as does activation of B cells (2). TH1 cells and NK cells also possess anti-tumorigenic activity (2). When taken together, an inflammatory tumor microenvironment influences tumor initiation, tumor heterogeneity and metastasis. In addition, the inflammatory tumor microenvironment can be modulated by therapeutics. Anti-inflammatory agents are thought to mitigate tumor growth by reducing inflammatory influences as opposed to cancer chemotherapeutic agents, which are thought to accentuate inflammation in the tumor microenvironment (1). In this latter case, conventional cytotoxic chemotherapeutic agents can cause massive, indiscriminate cell death leading to release of dead cell contents that increase inflammation and promote immunosuppression (1, 6).
The skin is an exemplar of the aforementioned inflammatory processes. It has a collection of disparate cell types, which work together to form an important innate immune barrier. However, many of these cell types can contribute to inflammation and, ultimately, to tumor formation. Keratinocytes, melanocytes, local dendritic cells, and Langerhans cells can trigger both the innate and adaptive immune systems via production and secretion of pro-inflammatory cytokines (2, 7, 8). Based on data from studies on ultraviolet B radiation (UVB) and keratinocytes, it appears that UVB induces damage that triggers caspase-1 activation and inflammasome formation in keratinocytes leading to apoptosis (9). Indeed, there is support for a direct link between inflammation in the skin, inflammasome formation, and increased NF-ĸB activity, with tumor formation (2).
Malignancies of the skin can be classified as basal cell carcinoma (BCC), squamous cell carcinoma (SCC), melanoma, and nonepithelial skin cancer (4, 10). It has been reported that SCC is more inflammatory compared to BCC (2, 11). Both SCC and BCC have been shown to have high levels of IL-17 and IL-22 secreting T cells that may help drive tumor growth (2).
However, there is a need for reliable methods to assess inflammatory infiltration in intact skin and elucidate the role of the immune system and immunosuppressants in oncogenesis and progression of SCC and BCC. Here, we present our findings supporting this method together with observations on the differences between SCC and BCC in kidney transplant patients treated with immunosuppressants compared to un-transplanted and un-treated patients.
Clinical characterization of kindey transplanted (KT) and non-transplanted (non-KT) patients. Data represent mean±sd.
Patients and Methods
This retrospective, observational study was performed in accordance with the ethics review board of the “Ruggi” Hospital and University of Campania “L. Vanvitelli”. The study included kidney transplanted (KT) patients with a histological diagnosis of either BCC or SCC. A control group was selected consisting of non-transplanted patients with normal kidney function who had a histological diagnosis of BCC or SCC, which were age-matched with KT patients.
The study included only patients with available biopsy specimens stained with Hematoxylin-Eosin (see below). In addition, the biopsy specimen had to include a region of non-cancerous intact skin region, without evident histological alterations (i.e. cellular and nuclear atypia).
Exclusion criteria for control subjects were the presence of immunosuppressive states including the use of anti-rejection drugs, genetic predisposition to tumors, and chronic inflammatory diseases. Exclusion criteria for KT were advanced chronic kidney disease (stage IV-V). Clinical data are reported in Table I.
Histological processing. The skin excisions were formalin-fixed, paraffin-embedded and processed according to standardized procedures. Sections were stained with standard histological stains (H&E, trichrome). Only H&E staining was further considered in the present work. Diagnosis of BCC and SCC was performed by expert pathologists (AC, AR, PZ, RF).
Image analysis. Images of the intact skin in the biopsy sample were digitally acquired with a Zeiss microscope (objective 20×). Several images were acquired to cover the entire region of the intact skin (in some cases up to 1 mm), while ensuring an overlap between the images. Subsequently the entire skin (whole slide images, WSI) was reconstructed using the Image Composite Editor 2.0.3.0 64bit by Microsoft. The reconstructed wide fields were then quantitatively analyzed using the ImageJ image analysis software using a validated algorithm described elsewhere (12).
RGB images were first split in monochrome channels (using the “split channel” tool). The green (cell nuclei) and the blue channels (cytoplasm) were then inverted and their difference calculated using the imageCalculator tool. The green-blue difference was used for the analysis because this operation resulted in a greater signal to noise (S/N) ratio of the nuclei.
Results from quantitative analysis of intact skin properties of kidney transplanted (KT) patients and patients without transplant, diagnosed with skin cancer. Data represent mean±SD.
In the resulting images the highest pixel values corresponded to greater intensity of the nuclear staining. These were then analyzed to identify the number of nuclei in the basal cell layer.
The following variables were quantified: number of nuclei in the basal cell layer (BCL) per unit length, number of mitoses in BCL per unit length, number of inflammatory and stromal nuclei adjacent to the BCL per unit length. These data were measured by observers who were blinded to the patient group.
To obtain quantitative estimates of the nuclear distribution and number, a semi-automatic strategy was pursued: a segmented line with a thickness adjusted to cover the height of the basal cell layer (8 microns) was manually drawn onto the basal cell layer. The “plot profile” tool was then used to measure the average nuclear signal intensity along the line (intensity profile). An estimate of the number of nuclei per unit length was then obtained by counting the number of peaks on the intensity profile. The same signal was also analyzed using the Fourier analysis to identify the main spectral component in the spatial distribution of nuclei. Mitotic cells were identified by their greater signal intensity (which was due to the doubling of the amount of DNA) by counting the peaks measuring two standard deviations above the average peak intensity.
To measure the regularity or randomness of the inter-nuclear spacing, we measured the entropy and fractal dimension of the intensity profile. The variability of nuclear distribution was also measured by measuring the standard deviation of the nuclear-to-nuclear distances over 10 nuclei (SDNN10).
To measure the number of inflammatory cells, a ribbon of constant thickness was centered onto the basal cell layer of the tumor using the “segmented line” tool and adjusting its thickness. The ribbon was then straightened using the “straighten” tool in ImageJ. Afterward, a rectangular ROI from the end of the basal cell layer and to the adjacent connective tissue was selected and the number of nuclei were counted using a manually selected threshold and the “Analyze particles” tool. Total inflammatory cells were then expressed as cells per unit length (referring to the length of the basal cell layer).
Statistics. Data were analyzed using the R environment. Differences between the KT and non-KT control group regarding the clinical variables were tested using Student's t-test without the assumption of homogeneity of variance (Welsh method). A two-way ANOVA using presence of kidney transplant (KT vs. non-KT) and cancer type (BCC vs. SCC) as factors was used to verify the effect of kidney transplantation and of the type of skin cancer developed onto intact skin microarchitecture. Data are reported as mean±standard deviation. The statistical threshold for significance was set at p<0.05.
Results
Using the whole-slide reconstruction approach, intact skin images measuring up to 3 mm in length were analyzed, with total nuclear counts up to 432 nuclei. Cumulatively, 10,039 nuclei in the basal cell layer and 4,738 nuclei in the adjacent connective tissue were examined in this study. The two study groups were comparable for most clinical parameters, as reported in Table I, except for the renal function, which was reduced in KT patients (indexed by creatinine, urea, eGFR). Lower levels of hemoglobin were also noted, albeit the clinical impact of this difference was negligible. The total ratio of cell nuclei in the basal cell layer and the nuclei in the adjacent connective tissue was 2:1, where two epithelial cells in the basal cell layer corresponded to 1 cell in the subepithelial connective tissue using the line thickness adopted in the study. To compare the non-linear variables from skin samples of different lengths, we used a sliding window approach: all data were cut into pieces of 250 micron, and then the average of these pieces was considered for each variable. This allowed for a very precise estimate of the variables reported in Table II.
When considering the entropy of the spatial distribution and the standard deviation of the nuclear distances (SDNN10), no difference was noted in the basal cell layer of KT patients compared to non-KT controls. However, cells were more densely packed (greater spatial frequency and greater linear density) in the basal cell layer of KT patients who developed a BCC, compared to non-KT patients (Table II and Figure 1). Conversely, KT patients with a diagnosis of SCC showed reduced density of cells in the basal cell layer. This interaction effect is reported in Table II and Figure 1.
Modifications in skin architecture linked to kidney transplant and skin tumor type. Left panel: the density of nuclei in the basal cell layer depends on the type of tumor: in subjects with SCC the immunosuppressive regimen reduces cell density, whereas in patients that developed BCC the same drugs increased nuclear density in the intact skin. Right panel: kidney transplant is accompanied by an increase in the number of nuclei close to the basal cell layer, including inflammatory cells, compared to non-KT patients, regardless the nature of the tumor that has then developed.
This difference was not accompanied by a corresponding significant change in the number of mitotic figures. Furthermore, KT was accompanied by an increase in the number of nuclei close to the basal cell layer, including inflammatory cells, compared to non-KT patients, regardless of the nature of the tumor these subjects developed (Table II, Figure 1).
Thus, when taken together, the results presented here support our prior report on using whole slide reconstruction to analyze the inflammatory architecture of BCC and reveal divergent characteristics for SCC which open the door to further studies.
Discussion
Our results reveal important features of SCC and BCC in kidney transplant patients and validate our non-linear, microarchitecture image analysis technique as a tool for studying inflammation in biopsy specimens. While other reports have previously summarized the presence of non-malignant skin conditions in kidney transplant, this is the first report to quantitatively analyze the modifications of the skin linked to transplantation (13).
The main result of the present study is that KT patients who developed BCC or SCC showed an increased number of inflammatory cells in the intact skin, compared to non-KT patients. In a previous study focusing on BCC, we noticed a greater inflammatory infiltrate in KT patients (12). The present work confirms this observation and extends it to KT patients with SCC. Because of the different kidney function (indexed by the eGFR) between non-KT and KT patients, it is not possible to ascertain whether skin alterations are driven by chronic kidney disease (CKD) or by the immunosuppressive regimen. However, we discuss below these possibilities using correlation analysis. Interestingly, the density of cells in the basal cell layer differed markedly, and paradoxically, between SCC and BCC (Figure 1). As noted in the results, we observed increased inflammatory infiltrates in BCC patients who received a kidney transplant and immunosuppressive agents but decreased inflammatory infiltrates in SCC patients who received a kidney transplant and immunosuppressive agents. However, prior studies have revealed that there is a significant difference in the number of white blood cells (WBCs) (including neutrophils and monocytes) between BCC and SCC. Indeed, WBCs are decreased in BCC compared to SCC and the ratio of neutrophils:leukocytes is lower in BCC compared to SCC (3.24:3.59). The interaction effect between transplant and tumor type regarding the density of cells in the basal cell layer is particularly hard to interpret and additional studies with appropriate control groups (i.e. non-KT CKD patients with BCC or SCC) are needed to disentangle this point.
At the present time, the underlying reasons for the differences observed in nuclear density and inflammatory cell density remain unclear. However, there are several potential possibilities:
It is unrelated to cancer, and the increased number of nuclei in the connective tissue is due to fibroblasts and not inflammatory cells. This possibility should be analyzed when considering that a small portion of patients undergoing therapy with calcineurin inhibitors (CNI, cyclosporine or tacrolimus) may show gingival hyperplasia and hypertricosis. However, this interpretation is unlikely to occur because of the different shape of the inflammatory cells compared to fibroblasts in the connective tissue. Furthermore, the low dose of steroids universally used in KT is expected to reduce skin thickness rather than increase it. Interestingly, cyclosporine has been shown to increase sonic hedgehog (SHH) signaling that, in part, drives gingival hyperplasia (14, 15). Hedgehog signaling is also a key player in mediating the inflammatory response in BCC (2). This suggests, a priori, that treatment with cyclosporine, in the context of BCC, may lead to an augmented inflammatory response that could help explain the observations reported in this study. However, further confirmatory analysis using immunohistochemical techniques is required to better characterize the phenotype of the inflammatory infiltrate. Gene expression analysis may also lead to a greater insight into the nature of the inflammatory process.
It is unrelated to cancer and is due to a basal inflammatory state caused by chronic rejection or CKD. This possibility stems from the proposal that CKD represents a pro-inflammatory state (16), although no data are available concerning skin inflammatory cells. Furthermore, a certain grade of low-grade chronic rejection is likely present in all KT patients. A report in 2011, based on a swine model of KT, showed a linear relation between kidney graft rejection (indexed by Banff scoring system) and histological changes (vacuolar alteration) in the epidermis (17). Unfortunately, the authors did not examine the histopathological changes in the subepithelial connective tissue. This condition has been described by the same group also in the case of intestinal transplantation (swine-model) (18). The authors consider this as an atypical form of “cutaneous graft-versus-host disease” (19), which usually follows allogenic hematopoietic stem cell transplantation and not kidney KT. In human subjects, cases of graft-versus-host disease in kidney transplant (20, 21) and in liver transplant (22) have been described, with clear cutaneous reactions (e.g. rash), but these are exceptional cases. We did not find any correlation between the density of subepithelial inflammatory cells and eGFR or proteinuria, as indices of kidney damage. Furthermore, there was no correlation between white blood cell counts (as index of inflammation) and density of subepithelial inflammatory cells. Therefore, it is unlikely that these changes reflect a general inflammatory state due to CKD or rejection. However, the authors recognize that it is appealing to use skin biopsies for the identification of the rejection state.
It is an effect of cancer, possibly a consequence of an inflammatory escape. An inflammatory infiltrate is often present in many solid malignancies, unfortunately without the ability to overcome cancer growth. The amount of inflammatory infiltrate is different among tumors: in our case SCC is usually accompanied by a stronger peritumoral inflammatory reaction. However, we found similar quantitative response in the skin from SCC and BCC, without tumor type effect, which would argue against an unspecific cancer-driven effect.
It is a cause of cancer, a general precancerous condition induced by the immunosuppressive state. This interpretation stems from the known greater risk for skin cancer in transplanted patients. KT patients show an increased risk for all skin tumor types, with slightly greater prevalence of SCC compared to BCC (i.e. BCC:SCC ratio of 1:4), which is the opposite of the general population (23). The immunosuppressive regimen would be expected to reduce the inflammatory infiltrate, as evidenced by the reduced prevalence of inflammatory skin lesions in KT (24). Steroids are expected to decrease skin thickness. CNI inhibitors are known to induce excessive hair growth, acne, and, rarely, gum hyperplasia (25). Therefore, our findings might be specific to the transplanted patients in which transplantation led to the development of a skin malignancy. Unfortunately, the study was not designed to prove this hypothesis, and additional control groups are needed in future studies (i.e. transplanted patients without cancer).
Furthermore, additional studies are needed to determine the role of immunosuppressants or cancer disease process in mediating inflammation in SCC and BCC.
Footnotes
Authors' Contributions
AnC, DV, MWL, GC provided the first draft of the manuscript. AnC, DV, GB, MS, GC provided the statistical analysis of the data. GP, GB, CS, RP recruited and analyzed patients; DV, AR, AnC, CS, AlC, PZ, RF provided the image analysis. All Authors modified the final draft of the manuscript.
Conflicts of Interest
The Authors declare no conflicts of interest in relation to this study.
- Received May 5, 2020.
- Revision received May 21, 2020.
- Accepted May 23, 2020.
- Copyright© 2020, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved
