eTable 1. Frequency of Itch and Pain in All Cancer Types
eTable 2. Frequency of Itch and Pain by Neoplasm Type
eTable 3. Comparison of Itch VAS Scores for Lesions That Report Itch and Pain to Those With Itch Only
eFigure 1. Consort Flow Diagram
eFigure 2. Representative Images of Inflammation Type With Cancer Type
Yosipovitch G, Mills KC, Nattkemper LA, Feneran A, Tey HL, Lowenthal BM, Pearce DJ, Williford PM, Sangueza OP, D’Agostino RB. Association of Pain and Itch With Depth of Invasion and Inflammatory Cell Constitution in Skin CancerResults of a Large Clinicopathologic Study. JAMA Dermatol. 2014;150(11):1160-1166. doi:10.1001/jamadermatol.2014.895
This study highlights a simple bedside evaluation of itch and pain for suspicious skin lesions.
To examine the correlation of pain and itch with histologic features of skin cancers.
Design, Setting, and Participants
This large, prospective, clinicopathologic study enrolled patients who filled out questionnaires that assessed itch and pain intensity of their skin tumors at the time of excision. Study participants were from the patient population presenting to the Department of Dermatology surgical unit at Wake Forest University Baptist Medical Center from July 1, 2010, through March 31, 2011. Study participants included 268 patients, representing 339 histopathologically confirmed cutaneous neoplasms. The following skin cancer subtypes were represented in this analysis: 166 basal cell carcinomas, 146 squamous cell carcinomas, and 27 melanomas.
Main Outcomes and Measures
Itch and pain associated with skin cancer at the time of excision ranked on an 11-point (score range, 0-10) numerical visual analog scale and histopathologic analysis for each neoplasm (assessment of the amount and type of inflammation, ulceration, perineural invasion, and depth of invasion).
The prevalence of itch and pain across all skin cancers was 36.9% and 28.2%, respectively. However, these symptoms were mostly absent in melanomas. Pain intensity was significantly associated with the degree of inflammation (mild or none vs moderate or marked; P < .001), presence of neutrophils in the inflammatory infiltrate (predominantly mononuclear vs mixed or neutrophilic; P = .003), presence of eosinophils (present vs absent; P = .007), ulceration (yes vs no; P = .003), perineural invasion (yes vs no; P < .001), depth of invasion (P = .001), and largest diameter length of skin lesion (P < .003). Itch intensity was significantly associated with the degree of inflammation (mild or none vs moderate or marked; P = .001) and the presence of eosinophils (present vs absent; P = .02).
Conclusions and Relevance
These findings support the theory that itch emanates from the upper layers of the skin, whereas pain is associated with deeper processes. This study also reports that a simple bedside assessment for the presence and intensity of pain or itch is an easily implementable tool for physicians evaluating suspicious skin lesions.
Skin cancer is the most common form of cancer in the United States, with an incidence that has been steadily increasing.1 Although skin cancer is common, data on frequently encountered dermatologic symptoms associated with these cancers are still emerging. A large prospective study2 reported in 2007 found itch to be a common feature of nonmelanoma skin cancer. This study2 also reported tenderness to be more frequently associated with squamous cell carcinoma (SCC) compared with basal cell carcinoma (BCC) or melanoma. Similarly, Mills et al reported findings on a prospective study that examined the prevalence and intensity of itch and pain in nonmelanoma skin cancer. The study3 confirmed pain to be a more common feature of SCC compared with BCC. However, to our knowledge, no study has examined the association between histologic features of common skin cancers with itch and pain, 2 common dermatologic symptoms with validated rating scales.4,5 Thus, the goal of the present study was to examine the correlation of itch and pain with histologic features of skin cancers.
This analysis was part of a larger study on skin cancers that involved 603 total cutaneous neoplasms from 505 different patients (eFigure 1 in the Supplement). Only cutaneous neoplasms from skin biopsy specimens available for histopathologic examination were included in the present study. The study was designed to examine the correlation of itch and pain with cancer type, subtype, tumor size, and several histologic features, including the amount and type of inflammation, perineural invasion, ulceration, and depth of invasion. The Wake Forest University School of Medicine Institutional Review Board approved this study. Participants had biopsy-proven BCC, SCC, malignant melanoma, or melanoma in situ and were recruited from the patient population presenting to the Department of Dermatology surgical unit at Wake Forest University Baptist Medical Center from July 1, 2010, through March 31, 2011. Under guidance from a trained member of the research team (A.F. or Alex Zeitany, BA), the participants signed a written informed consent form and were then asked to fill out a questionnaire designed to identify their symptoms of itch and pain associated with their tumor. The study participants were asked to rank their perceived sensations of itch and pain on an 11-point (score range, 0-10) numerical visual analog scale (VAS). The VAS is anchored with the verbal descriptors of “no sensation” on the left and “the most intense sensation imaginable” on the right.
Slides from the archives in the Department of Pathology of Wake Forest University Baptist Medical Center were retrieved for a total of 339 neoplasms from 268 different patients who participated in the larger study. The pathologist (K.C.M. or O.P.S.) who reviewed the slides was masked to all patient data, including the VAS scores linked to each cancer. Slides were first examined for confirmation of the previously rendered diagnosis and confirmation of subtype. Histopathologic analysis for each neoplasm was also performed for assessment of the amount and type of inflammation, ulceration, perineural invasion, and depth of invasion. The analysis only includes data from the lesions for which the original diagnostic shave biopsy material was available for review, and only the slides that contained the original diagnostic tissue were assessed.
For each of the histologic parameters used in the analysis, a specific set of guidelines was followed to standardize the assessment of each feature (eFigure 2 in the Supplement). Ulceration was scored as present when an ulcer bed was identified overlying the neoplastic cells. Perineural invasion was scored as present if neoplastic cells could be visualized invading a nerve fiber or if neoplastic cells were identified beneath the perineurium adjacent to the nerve fiber. The amount of inflammation was stratified into the following 4 categories: (1) marked inflammation in which more than 50% of the nonelastotic dermis directly adjacent to the neoplastic process was occupied by inflammatory cells in at least 2 high-power fields (HPFs) (original magnification ×400 with a field diameter of 0.5 mm), (2) moderate inflammation in which 10% to 50% of the nonelastotic dermis directly adjacent to the neoplastic process was occupied by inflammatory cells in at least 2 HPFs (original magnification ×400), (3) mild inflammation in which less than 10% of the nonelastotic dermis directly adjacent to the neoplastic process was occupied by an inflammatory infiltrate, and (4) no inflammation in which there was no perceptible inflammatory infiltrate. If the biopsy specimen was so small that fewer than 2 HPFs of nonelastotic dermis were present, then the samples were rendered as unable to assess.
Inflammation type was stratified into the following 3 categories: (1) significantly neutrophilic in which more than 5% of the inflammatory infiltrate consisted of neutrophils (polymorphonuclear leukocytes [PMNs]) as identified in the nonelastotic dermis directly adjacent to the neoplastic cells in 5 HPFs, (2) mixed type in which PMNs were present but less than 5% of the inflammatory infiltrate consisted of PMNs, and (3) nonneutrophilic in which the inflammatory infiltrate had no definitively identifiable PMNs in the nonelastotic dermis adjacent to the neoplasia.
The presence of eosinophils was recorded in a separate category independent of the inflammation type category. Eosinophils were scored as being present if at least 1 eosinophil was present in a 5-HPF assessment in the dermal interstitium directly adjacent to the neoplasia.
The depth of invasion was assessed by measuring the distance from the top of the granular layer of the epidermis to the deepest extent of dermal infiltration by the tumor using a vernier scale.6 If the lesion was ulcerated, the base of the ulcer was used as the upper point of reference. In the invasive malignant melanomas, the Breslow thickness of the tumor as designated on the initial pathology report was used as the depth of invasion.
Descriptive statistics were calculated for categorical variables using numbers and percentages and for continuous variables using means and SEs because the data collected on a continuous scale had an approximately normal distribution. There were 7 primary variables that were used in analyses: depth of invasion, ulceration, degree of inflammation, presence of neutrophils in the inflammatory infiltrate, presence of eosinophils, perineural invasion, and largest diameter length of skin lesion. For each of the primary variables, analyses were performed that examined whether there were associations with pain and itch intensities overall and then separately by skin cancer type (SCC and BCC). For binary variables (ie, presence or absence of eosinophils), 2-sample t tests were used to compare mean pain and itch VAS scores. Linear mixed models were fit to determine the effect of each primary variable on the outcome of pain and itch, adjusting for age, sex, presence of chronic pain or itch-related conditions, and transplantation status. These models adjusted for multiple lesions that occurred on the same patient by including the patient as a random effect in the model. General linear models were performed on a per-lesion (rather than per-person) level to compare with the mixed models to determine whether there was evidence of a high within-person correlation for pain and itch scores.
Pain and itch scores were also dichotomized (present or absent), and Fisher exact tests, with odds ratios and 95% CIs, were calculated when examining each binary primary variable (ie, eosinophils) with the pain and itch variables. Multiple logistic regression models were also fit to examine these associations, adjusting for age, sex, presence of chronic pain or itch-related conditions, and transplantation status. An additional analysis that adjusted for patients with multiple lesions was performed. For analyses in which the key predictor had more than 2 levels (ie, VAS symptom score broken into 3 categories: 0, 1-4, and 5-10), an overall adjusted model was first examined using F tests, and then pairwise comparisons among the different categories were made using 2-sample t tests. The analyses were performed on the full sample and then stratified by skin cancer type (BCC and SCC). For these analyses, 95% CIs were calculated as well.
This study consisted of 268 patients, representing a total of 339 histopathologically confirmed cutaneous neoplasms. The following skin cancer subtypes were represented in this analysis: 166 BCCs, 146 SCCs, and 27 melanomas (including invasive melanoma and melanoma in situ). The prevalence of itch and pain across all skin cancers was 36.9% and 28.2%, respectively (Table 1). Quiz Ref IDThe prevalence of itch was greatest in SCC (46.6%), followed by BCC (31.9%) and melanoma (14.8%). Pain prevalence was also greatest in SCC (42.5%), again followed by BCC (19.9%) and melanoma (3.7%). The prevalence of itch and pain was significantly greater in SCCs (P = .002) and BCCs (P < .001) compared with melanoma. Additional patient demographic information, symptom characteristics, and histologic features based on neoplasm type are presented in Table 1 and Table 2.
In addition, several lesions were reported to have pain and itch. The prevalence of lesions with itch that reportedly also had pain was 45.6%, and the prevalence of painful lesions that reportedly also had itch was 60.0% (eTable 1 in the Supplement). The prevalence of itch in painful BCC lesions was 60.7%, whereas the prevalence of pain in itchy BCC lesions was 37.7%. In SCC cases, 52.9% of lesions with itch reportedly also had pain, and 59.0% of those with pain reportedly also had itch. Of 4 itchy melanoma lesions, only 1 had pain, and only 1 painful lesion had itch (eTable 2 in the Supplement). Furthermore, the overall mean (SE) VAS score for itch intensity for lesions that exhibited itch and pain (3.93 [2.12]) does not significantly differ from the VAS score of those with only itch (3.79 [1.98]) in all cancer types or each neoplasm type (eTable 3 in the Supplement).
When all skin cancers were grouped, the mean depth of invasion was highest in the most painful lesions (VAS score of 5-10; mean [SE] depth of invasion, 0.17 [0.05] cm), followed by intermediately painful lesions (VAS score of 1-4; mean [SE] depth of invasion, 0.12 [0.01] cm), and was lowest in the painless lesions (VAS score of 0; mean [SE] depth of invasion, 0.1 [0.05 and 0.02]). Quiz Ref IDStatistical significance was achieved when comparing depth of invasion among the most painful and painless lesions (P = .001). When analyzed individually, this finding was significant (P = .05) for SCC samples but not for BCC samples (P = .07). Depth of invasion did not correlate with intensity of pain in melanoma samples.
For itchy lesions, when all skin cancers were grouped, the mean (SE) depth of invasion was 0.14 (0.04) cm in the itchiest lesions (VAS scores of 5-10), 0.12 (0.02) cm in the intermediately itchy lesions (VAS scores of 1-4), and 0.1 (0.005) cm in nonitchy lesions (VAS score of 0). Depth of invasion was not significantly associated with itch when the itchiest and nonitchy skin lesions were compared in all cancers (P = .09), as well as in individual analyses of SCCs (P = .31) and BCCs (P = .67). Depth of invasion did not correlate with intensity of itch in melanoma skin lesions.
When all skin cancers were grouped using the linear mixed-model approach, cancers that were ulcerated had significantly (P = .003) higher mean (SE) VAS scores for pain intensity (2.88 [0.37]) compared with those that were nonulcerated (2.02 [0.29]). This association was found also in patients with BCC (P = .01) and showed a similar association in SCC (VAS score of 3.27 in patients with ulcerated skin cancers vs 2.37 in patients without ulcerated skin cancers), but this finding did not reach statistical significance (P = .10).
When examining the association between itch and ulceration using the linear mixed-model approach, we found no significant results (mean [SE] VAS scores of 1.88 [0.43] for those with ulcerated skin cancers vs 2.01 [0.34] for those without ulcerated skin cancers; P = .87). No association was found between ulceration and itch in patients with BCC. Interestingly, on analysis of SCCs, a trend of an inverse association was found between the presence of ulceration and itch intensity. The mean (SE) VAS score was 1.55 (0.66) for itch intensity among ulcerated skin lesions and 2.65 (0.44) for nonulcerated skin lesions (P = .06). None of the melanoma samples in our analysis exhibited ulceration; thus, the association with pain or itch could not be examined.
The presence of pain significantly increased the odds of skin cancers exhibiting a marked or moderate degree of inflammation compared with mild or no inflammation (Table 3). Regarding intensity, pain intensity of all skin cancers with marked or moderate inflammation was significantly greater than those with mild or no inflammation (P < .001) (Table 4). Analysis of BCCs revealed similar findings (P = .01) but not in SCCs (P = .10).
The presence of itch also significantly increased the odds of skin cancers exhibiting a marked or moderate degree of inflammation compared with mild or no inflammation (Table 3). Regarding intensity, the mean VAS score for itch intensity was significantly higher in skin cancers with marked or moderate inflammation compared with mild or no inflammation (P < .001) (Table 5). Analysis of BCCs also revealed a strong association between degree of inflammation and itch intensity (P = .01), but SCCs did not have an association (P = .21). An analysis of the association between pain and itch and inflammatory infiltration was limited in melanoma samples because of an insufficient number of patients exhibiting symptoms.
The presence of neutrophils in the inflammatory infiltrate was associated with an increased prevalence of pain (P = .002) (Table 3) and increased pain intensity (P = .003) (Table 4). A stronger association between pain and the presence of neutrophils in the inflammatory infiltrate was seen in SCCs compared with BCCs.
The presence of neutrophils in the inflammatory infiltrate was also associated with an increased prevalence of itch (P = .06) (Table 3) and increased itch intensity (P = .07) (Table 5). Analysis of different cancer types reveals no association with itch intensity. An analysis of the association between pain and itch and the presence of neutrophils in the inflammatory infiltrate could not be performed in melanoma because neutrophils were not detected in the inflammatory infiltrate in any of the melanoma samples.
The presence of eosinophils was significantly associated with an increased prevalence of itch (P = .02) (Table 3) but not prevalence of pain (P = .09) (Table 3). The presence of eosinophils was also associated with increased pain (P = .007) and itch (P = .02) intensities (Table 4 and Table 5). The association of eosinophils with pain and itch was stronger in SCCs compared with BCCs. An analysis of the association between pain and itch and the presence of eosinophils in melanoma was not performed because eosinophils were not detected in any of the melanoma samples.
Only 3 skin cancers featured perineural invasion (2 SCCs and 1 BCC). These samples were noted to have a high pain intensity rating (mean [SE] VAS score for pain of 8.17 [1.13] compared with 2.09 [0.27] for other samples [P < .001]) and VAS score for itch of 3.14 [1.31] compared with 1.99 [0.31] for other samples [P < .001]).
Quiz Ref IDAn analysis of the association between the largest diameter length of skin lesions and itch or pain revealed a clear association between larger diameter length and increasing pain and itch intensity. In particular, the most painful lesions (VAS score ≥5) had a mean diameter of 2.19 cm (95% CI, 1.82-2.55 cm) compared with a mean diameter of 1.59 (95% CI, 1.45-1.74) in painless lesions (VAS score of 0) (P = .003). The itchiest lesions (VAS score ≥5) had a mean diameter of 2.04 cm (95% CI, 1.72-2.36 cm) compared with a mean diameter of 1.55 (95% CI, 1.40-1.69) in lesions that were not itchy (VAS score of 0) (P = .005).
An analysis of the various subtypes of BCCs and SCCs revealed a statistically significant increase in the mean (SE) pain ratings only in the morpheaform type of BCC (VAS for pain of 3.40 [0.66]) compared with all other BCCs (VAS score for pain of 1.59 [0.35]) (P = .01).
Additional analyses were performed using multiple logistic regression analyses to take into account the presence of multiple lesions on certain patients. These analyses revealed no significant differences from the models that were performed on a per-lesion level.
This prospective clinicopathologic study with a large data set found that pain and itch are associated with certain histopathologic characteristics. We examined depth of invasion, which has been previously reported to be a prognostic marker for survival in cancer.7- 9 Significantly, the intensity of pain was associated with depth of invasion. The presence of ulceration was also associated with pain intensity, whereas itch intensity was not significantly associated with ulceration or depth of invasion.
It has been previously reported in only melanoma that depth of invasion (increased Breslow depth) was related to the presence of pain and itch (odds ratios, 3.3 and 1.9, respectively).10Quiz Ref IDOur study did not find any association with pain and itch to depth of melanoma, and our clinical experience does not suggest that pain and itch are frequent symptoms in patients with melanoma lesions. However, in nonmelanoma skin cancers, these are common symptoms. Future studies should further assess pain and itch in patients with melanoma to draw firm conclusions.
Our finding supports the theory that itch emanates from the upper layers of the skin, whereas pain is often the result of deeper processes. Ulcerated lesions, which involve deeper processes in the dermis, were associated with pain and not itch. Further support for this theory is provided from the pivotal study by Shelley and Arthur11 that found that the removal of papillary dermis induces pain and not itch.
The statistical modeling performed allowed for a comparison of results based on a per-lesion model (ie, each lesion treated as an independent observation in the models) and a per-patient model (ie, patients with multiple lesions were appropriately modeled using a linear mixed-model approach). In these analyses, we found little difference between results that were performed on a per-patient or per-lesion basis, suggesting that each lesion on each patient operates independently of other lesions on that patient. In other words, patients who had multiple lesions were less likely to have high or low pain and itch scores than patients with 1 lesion.
Quiz Ref IDThe findings that greater amounts of inflammation correlated with increased intensities of pain and itch are not surprising because many pruritogenic and nociceptive mediators are released with increased inflammation. In regard to specific inflammatory cell constituents, the presence of neutrophils appears to be a key driver of the pain response. Several neutrophilic conditions are associated with painful conditions, including Sweet syndrome, lobular panniculitis, and pyoderma gangrenosum.
The results of our study also reveal eosinophils to be associated with itch and pain. The association of eosinophils has been well documented to be associated with atopic eczema,12 hypereosinophilic syndrome,13 and prurigo nodularis.14 A mechanism for eosinophilia that induces pruritogenic and nociceptive symptoms may be found in a study15 that reported that eosinophils, but not mast cells, significantly increased sensory neuron branching in vitro and in vivo. The association of eosinophilia with pain is a more novel finding. Mechanistically, eosinophils are known to be linked to the storage and production of neurotrophins, including nerve growth factor and neurotrophin 3,16 which have been reported to be mediators of pain.17 Interestingly, a study18 of patients with cutaneous findings of Churg-Strauss syndrome, a condition that features eosinophilia, reported an association with mononeuropathy multiplex, a severely painful condition.
Our study is limited in that the population of patients had a larger number of men than women. Furthermore, patients in this study had an increased likelihood of having tumors on the face and scalp because patients were recruited from the outpatient dermatologic surgery clinic. Even though the itch- and pain-rating survey was administered the same week of the shave biopsy procedure for most patients, the data could have been biased by the procedure or memory retrieval capabilities. Another limitation is the relatively small sample size of melanoma cases, which does not allow us to draw firm conclusions about the association between pain and itch and histopathologic features of melanoma. However, it is clear that melanomas have a significantly decreased prevalence of pain and itch compared with nonmelanoma skin cancer, a finding confirmed by a similar study2 that examined itch and tenderness in patients with melanoma. However, our study highlights the importance of a simple bedside evaluation for the presence and intensity of pain and itch as an easily implementable tool for physicians to use when evaluating suspicious skin lesions.
In summation, this study revealed that pain is associated with histologic features that involve deeper dermal processes in SCC lesions, such as ulceration and depth of invasion, whereas itch is linked with the more superficial BCC lesions. Moreover, the prevalence of pain and itch increased the odds of all lesions having marked or moderate degrees of inflammation. Furthermore, the presence of neutrophils and eosinophils in the inflammatory infiltrate increased with higher pain and itch intensities. In conclusion, this study highlights the importance of the underlying histopathologic features of cutaneous neoplasms and their associations with itch and pain.
Accepted for Publication: April 8, 2014.
Corresponding Author: Gil Yosipovitch, MD, Department of Dermatology, Temple University School of Medicine, 3500 N Broad St, MERB 460, Philadelphia, PA 19140 (firstname.lastname@example.org).
Published Online: July 23, 2014. doi:10.1001/jamadermatol.2014.895.
Author Contributions: Drs Yosipovitch and D’Agostino had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Yosipovitch, Mills, Pearce, D’Agostino.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Yosipovitch, Mills, Nattkemper, D’Agostino.
Critical revision of the manuscript for important intellectual content: Yosipovitch, Mills, Nattkemper, Feneran, Tey, Lowenthal, Pearce, Williford, D’Agostino.
Statistical analysis: Mills, Lowenthal, D’Agostino.
Obtained funding: Yosipovitch.
Administrative, technical, or material support: Nattkemper, Lowenthal.
Study supervision: Yosipovitch, Pearce, Sangueza.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was supported in part by the Wake Forest Comprehensive Cancer Center (Dr Yosipovitch).
Role of the Sponsor: The funding source had no role in 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.
Additional Contributions: Shawn Kwatra, MD, Department of Dermatology, Wake Forest University Baptist Medical Center, Winston-Salem, North Carolina, assisted with the literature review and Alex Zeitany, BA, Department of Dermatology, Wake Forest University Baptist Medical Center, Winston-Salem, North Carolina, assisted with study recruitment.
Correction: This article was corrected on August 14, 2014, to fix an author’s name in the byline, Author Affiliations, and Author Contributions.