Malignant tumors that arise in chronic ulcerating scar tissue of the skin are known as Marjolin's ulcers.1 The scar is typically a burn scar and the cancer a squamous cell carcinoma, but the cancer can also be a basal cell carcinoma, melanoma, or sarcoma. This phenomenon has been described in a recent review, including 146 articles with case reports and 28 case series,2 but the risk for malignant degeneration of burn scars has not been estimated in an epidemiologic study.
Marjolin ulcer has an acute form with cancer occurring within 1 year of the injury and a chronic form with cancer occurring more than 1 year after the injury.1 The time from burn to cancer is often reported to be long with an average latency of 31 years.2 It has been claimed that the latency is inversely proportional to the patient's age at the time of the burn.3 Several mechanisms for the malignant degeneration have been proposed, including production of toxins in burned skin, constant irritation due to prolonged healing, and repeated damage leading to chronic inflammation.4 Chronic inflammation, whether induced by biologic, chemical, or physical factors, has been associated with increased risks for human cancer.5 Because prolonged healing of wounds might be a risk factor for the development of carcinoma, it has been suggested that effective wound care (eg, use of early skin grafting) could be important with respect to cancer prevention.6,7
Burns affect primarily the skin, but extensive burns also have substantial systemic effects. One such effect is suppression of the immune system, which may persist for a long time after a burn injury. Both cellular and humoral immunity appear to be affected with a subsequent high risk for infection.8 Other immunosuppressed persons such as patients undergoing organ transplants are at increased risk for certain types of cancer, especially nonmelanoma skin cancer and non-Hodgkin lymphoma.9,10
We estimated the risks for skin cancer and other cancers in a Danish population-based cohort of 16,903 patients with thermal or chemical burns that were severe enough to require in-patient hospital treatment. Patients were followed for up to 25 years from hospitalization to development of cancer.
The Hospital Discharge Register was used to identify patients with burns. The Register was established in 1977 and keeps information on approximately 99% of all admissions to nonpsychiatric hospitals in Denmark.11 The variables registered for each patient include the personal identification number, dates of admission and discharge, and up to 20 discharge diagnoses. The unique personal identification number that is assigned at birth to each resident of Denmark encodes information on sex and date of birth; it ensures unambiguous linkage of information between various registers. The diagnoses in the Hospital Discharge Register were coded according to a Danish version of the International Classification of Diseases, Eight Revision (ICD-8) from 1977 to 199312 and surgical procedures according to a Danish classification system (op-codes).13 In the Hospital Discharge Register, we identified 18,008 patients who had been discharged after a diagnosis of a thermal or a chemical burn (ICD-8 = 940–949) during the period 1978 through 1993. A fifth digit, which has been added to ICD-8 codes 940–949 in the Danish version of the classification, specifies the severity of the burn (first-degree: superficial, involving only the epidermis; second-degree: partial thickness, involving both the epidermis and dermis; and third-degree: full-thickness through the epidermis and dermis and into the fat). All patients were linked to the Central Population Register to verify the personal identification number and to obtain dates of death and emigration. We excluded 554 patients (3%) with an invalid identification number (mainly persons from Greenland, the Faroe Islands, and foreign countries) and 551 patients (3%) who had died during the hospitalization for burn. Thus, 16,903 persons remained in the cohort.
The cohort was linked to the Danish Cancer Registry to obtain information on cancer diagnoses. The Cancer Registry, established in 1943, is notified whenever a case of cancer is diagnosed and when changes are made to an initial diagnosis. The obligation to notify the Cancer Registry applies to all malignant neoplasms, including nonmelanoma skin cancer, as well as papillomas of the urinary tract and histologically benign tumors of the central nervous system. Tumors are coded according to a modified Danish version of the International Classification of Diseases, Seventh Revision (ICD-7)14 and in addition, since 1978, according to the International Classification of Diseases for Oncology (ICD-O-1).15 If a person develops more than one primary tumor, each tumor is entered and counted as an individual record; however, multiple tumors of the skin with identical morphologic characteristics (ie, the same first 3 digits of the ICD-O-1 morphology code) are recorded only once, even when they are located on different parts of the body. Such tumors are, however, allocated a specific code for being multicentric.
Follow up for cancer started on the first day of the month after the date of discharge for the injury and continued until the date of emigration or death, if applicable, or 31 December 2002, whichever occurred first. The number of cancers observed during follow up was compared with the number expected calculated from accumulated person-years and national cancer incidence rates for each sex and in 5-year age groups and 5-year calendar time periods. The standardized incidence ratio (SIR; the observed number of cancers divided by the expected number) was computed for all cancer and for individual cancers, with corresponding 95% confidence intervals (CIs). The SIR for individual cancers was based on the assumption that the observed number of cancer cases in a specific category follows a Poisson distribution.16 The confidence limits were calculated using Byar's approximation.17
The analysis for skin cancers included stratification on sex, age at injury, time since injury, and severity of injury (first-degree, second-degree, third-degree, unspecified). Supplementary analyses were made according to the site of the burn (head and neck, trunk, upper and lower extremities, airways, digestive tract, other and unspecified sites) yielding separate risk estimates for cancers arising at the site of the burn injury. Persons whose injury affected more than one site contributed person-years to all applicable categories. Finally, the analysis was stratified on whether skin transplantation was performed. Patients were categorized as having undergone skin transplantation if they had been skin grafted (op-code = 8996, 8998, 9000) during the same hospitalization as the burn or up to 3 months after discharge for the burn. Nonmelanoma skin cancers were divided into squamous cell carcinoma and basal cell carcinoma by use of the ICD-O morphology code, and the site of the skin tumor was determined by use of the ICD-O topography code.
The cohort consisted of 16,903 patients with thermal or chemical burns, 11,277 (67%) of whom were males and 5626 (33%) females (Table 1). A substantial proportion of the cohort (42%) consisted of children and teenagers under the age of 20 years. The number of discharged patients with these types of injuries declined gradually over time, from 1586 in 1978 to 763 in 1993. The most frequent site of burns was the extremities followed by the head and neck. Of lesions that were specified, 80% were thermal burn lesions and 20% were chemical burn lesions. Chemical burns occurred most frequently in the digestive tract (77%). Almost one fourth of the patients (22%) were treated with skin transplantation.
The cohort was followed for a total of 263,578 person-years, and the mean follow-up time was 15.6 years (range, >0–25 years). The overall number of observed cancer cases corresponded closely to the expected number, with a SIR of 0.99 (95% CI = 0.93–1.06) (Table 2). The risk for skin cancer was reduced by approximately 30%, reflecting reductions in the risks for nonmelanoma skin cancer (SIR = 0.7; 95% CI = 0.6–0.9) and malignant melanoma (SIR = 0.7; 95% CI = 0.4–1.1). When the cases of nonmelanoma skin cancer were grouped by histologic type, the SIR for squamous cell carcinoma was 0.9 (95% CI = 0.6–1.5) and the SIR for basal cell carcinoma was 0.7 (95% CI = 0.6–0.9).
The risks for malignant melanoma, squamous cell carcinoma, and basal cell carcinoma did not differ materially by the sex of the patient or the age at time of injury (Table 3). We saw small excesses of malignant melanoma and squamous cell carcinoma during the first year of follow up, with 3 observed cases for each tumor type. None of the 3 melanomas and only one of the 3 squamous cell carcinomas was, however, at the same site as the injury. The risk estimates were reduced in all subsequent periods of follow up. In a stratified risk analysis combining age at time of injury and latency, the risk pattern remained largely unchanged (not shown in the table).
We also conducted an analysis of skin cancers confined to the burned area of the body. SIRs were 0.3 for malignant melanoma, 1.2 for squamous cell carcinoma, and 0.7 for basal cell carcinoma for all burned sites combined (Table 3). The SIRs after injuries of the head and neck were 0.8 for squamous cell and 0.9 for basal cell carcinoma, whereas no malignant melanomas were seen after injuries at this site. Reliable risk estimates for cancers at other burned sites could not be made owing to very small numbers of observed cases. The risks for the 3 types of skin cancer did not differ materially by severity of the lesion or between persons with and without skin transplants. The risk for squamous cell carcinoma, however, tended to be somewhat lower among those with skin transplants than among those without.
We found no excess risk for sarcomas of all types combined (Table 2). Three sarcomas of the skin were observed, but only one might have occurred at the same site as the injury. Three sarcomas were derived from connective tissue, of which 2 might have occurred at the same site as the injury.
After excluding cancers of the skin, bone, and connective tissue, we saw a modest (6%) increase in the risk for all cancers combined (Table 2). This was due mainly to increased risks for cancers of the lung (SIR = 1.2), larynx (1.7), mediastinum (4.0), liver (1.6), and esophagus (1.5), which are sites strongly related to tobacco or alcohol use. None of the cancers of the lung, larynx, or mediastinum was observed in the subcohort of 137 persons with lesions of the airways (expected number = 1.3). One of 17 cases of esophageal cancer occurred among the 1403 persons with lesions of the digestive tract (expected number = 0.6). The number of cases of non-Hodgkin lymphoma observed was similar to that expected from rates in the general population (Table 2). Although there was no overall excess of this cancer, the risk was elevated in the subgroup of patients with third-degree lesions (SIR = 2.5; 95% CI = 1.2–4.5), on the basis of 11 cases, and a trend in risk was seen with increasing severity of the lesion (first-degree, SIR = 0.8 [95% CI = 0.2–2.2]; second-degree, SIR = 0.9 [95% CI = 0.5–1.6]; P for trend = 0.02).
We saw no indication of an increased risk for skin cancer in patients with a history of hospitalization for burns; on the contrary, we observed a relatively consistent pattern of reduced risks. This was true particularly for malignant melanoma and basal cell carcinoma; the relative risk for squamous cell carcinoma was close to unity. Even for the patients with the most severe lesions, there was no suggestion of an excess risk for skin cancer; the risk for non-Hodgkin lymphoma appeared to be increased in this subgroup, but the finding was based on small numbers. The entire cohort of patients with burn injuries showed increased risks for cancers of the lung, larynx, and mediastinum, but the excesses were not found in the subgroup of patients with affected airways. The slightly increased frequencies of cancers of the esophagus and liver support the view that a previous burn injury might be a weak marker of a lifestyle with a greater use of tobacco and alcohol than in the general Danish population.
The finding that burn patients are at no increased risk for skin cancer is surprising in light of the clinical descriptions of a tendency to malignant degeneration of burn scars, widely known as Marjolin ulcer.2 Marjolin ulcer is most commonly seen as squamous cell carcinoma, whereas basal cell skin cancer, malignant melanoma, and skin sarcoma have been described more infrequently. Although we covered a follow-up period of up to 25 years from the date of hospitalization for the injury, this might not have been long enough to capture the full spectrum of the malignant degeneration of burn scars in view of the previously reported average latency of 31 years.2 We did not, however, see a tendency to higher risks among the persons followed the longest (more than 10 years). According to one report,3 persons who were older at the time of a burn had shorter latencies between the burn and a cancer; however, we did not find such a risk pattern. We also did not detect any tendency to a clustering of skin cancers at previously injured sites. A likely interpretation could also be that a previous burn injury of the skin does not increase the rate at which a skin cancer is initiated, but rather facilitate tumor progression for cancers arising at a normal rate—perhaps assisted by compromised cell growth regulation of the affected tissue and reduced healing capability.
In case series, Marjolin's ulcers commonly involved the extremities, especially at sites predisposed to repeated trauma such as the flexor and extensor surfaces over joints and areas over bony prominences.6,7 We did not have detailed information on the sites of skin cancers, but there was no predilection for the injuries and skin cancers to co-occur on the upper or lower extremities. It has been suggested that repeated trauma and prolonged healing increase the risk for Marjolin ulcer.6,7 Because 22% of the patients with injuries in our study had received skin grafts, which is a higher proportion than reported in previous case reports and series,2 chronic or repeated ulceration might have been less frequent among our cohort members. In our results, the risk for squamous cell carcinoma was somewhat reduced among persons with transplants and slightly increased among those without.
Our study comprised a large unselected population of all patients hospitalized with burns in Denmark. Registration of subsequent cancer diagnoses in the Cancer Registry was independent of registration of the injury in the Hospital Discharge Register. Minor burn injuries treated at emergency rooms were not included in the study. However, patients are instructed to contact a plastic surgery department if the burn wound has not healed within 14 days, and they will then be admitted and receive a skin transplant. Thus, nonhealed burn injuries will usually lead to hospitalization. We therefore think that we have captured the patients that are of interest in respect to investigating the hypothesis of skin cancers arising in previously burned skin with a chronic ulcer.
Reporting of nonmelanoma skin cancer to the Cancer Registry might not be complete, but we have no reason to believe that the registration was less complete for cohort members than for the background population. In fact, because squamous cell carcinomas arising in burn scars are known to have a worse prognosis than sunlight-induced squamous cell carcinomas,6,7 the frequency of reports of this specific cancer is more likely to have been higher than usual. Because the topographic classifications of the injury and the skin cancer were quite broad, we were unable to determine whether the 2 lesions occurred on exactly the same area of the body. Also, skin cancers occurring at multiple sites had to be excluded from these analyses of site for burn injury and site of skin cancer, because the topographies for multiple skin cancers are not coded in the Danish Cancer Registry.
The reduced risk for skin cancer seen in our study may be due to avoidance of exposure to the sun because patients with burn scars often experience itching and pain when exposed to the sun18 and may not want to uncover their skin if the burn has disfigured them seriously. Physicians usually recommend that exposure to the sun be limited during the immediate recovery period to avoid scar hyperpigmentation.19 To our knowledge, long-term sun exposure habits after burn injuries have been investigated in only one previous study.20 This study included 91 Finnish patients with at least 5% total body surface area burns during childhood. When they were interviewed on average 17 years after the burn, 74% reported that they sunbathed regularly in the summer, and 78% reported that the scarred skin tanned only slightly or did not acquire tanning at all. Hypopigmentation after burn injuries is a well-known phenomenon that supposedly could lead to higher frequency of cancer. On the other hand, it was recently postulated that scar tissue has more ultraviolet B-absorbing substances than healthy skin21; these substances act like melanin and sunscreen. This suggestion was based on the observation that the synthesis of vitamin D was reduced when skin biopsy samples from children with burns were exposed to ultraviolet B light. If confirmed, this mechanism could also partly explain the observed reduction in skin cancer risk. It could also be speculated, however, that the different architecture in skin grafts, with lack of cell-rich deep layers of the dermis, could be associated with a reduced risk in these areas.
The risk for non-Hodgkin lymphoma was increased in the persons with the most severe lesions. If this is not a chance finding, it may be due to the immune suppression assumed to occur in patients with severe lesions, because increased risks for non-Hodgkin lymphoma have been seen in well-established groups of immunosuppressed patients, eg, patients with organ transplants9,10 and those infected with HIV.22 In the subgroup of patients with severe lesions, however, we saw no excess of nonmelanoma skin cancer, which is also reported to occur at increased rates among patients with organ transplants.9,10
In our data, the tendency for malignant degeneration of burn scars did not increase the risk of squamous cell carcinomas or other types of skin cancer above the level seen in the general population. On the contrary, a reduced risk was seen for basal cell carcinoma and melanoma. This pattern was evident even in patients with the most severe lesions in whom an excess of non-Hodgkin lymphoma was seen. It should be emphasized that these findings were based on a follow-up period up to 25 years, which may have been too short to fully explore the risk of Marjolin ulcer.
The authors thank Andrea Bautz at the Institute of Cancer Epidemiology, Danish Cancer Society, for computer assistance.
1. Treves N, Pack GT. The development of cancer in burn scars. An analysis and report of thirty-four cases. Surg Gynecol Obstet
2. Kowal-Vern A, Criswell BK. Burn scar neoplasms: a literature review and statistical analysis. Burns
3. Lawrence EA. Carcinoma arising in the scars of thermal burns. With special reference to the influence of the age at burn on the length of the induction period. Surg Gynecol Obstet
4. Trent JT, Kirsner RS. Wounds and malignancy. Adv Skin Wound Care
5. O'Byrne KJ, Dalgleish AG. Chronic immune activation and inflammation as the cause of malignancy. Br J Cancer
6. Edwards MJ, Hirsch RM, Broadwater R, et al. Squamous cell carcinoma arising in previously burned or irradiated skin. Arch Surg
7. Ozek C, Cankayali R, Bilkay U, et al. Marjolin's ulcers arising in burn scars. J Burn Care Rehabil
8. Heideman M, Bengtsson A. The immunologic response to thermal injury. World J Surg
9. Adami J, Gäbel H, Lindelöf B, et al. Cancer risk following organ transplantation: a nationwide cohort study in Sweden. Br J Cancer
10. Birkeland SA, Storm HH, Lamm LU, et al. Cancer risk after renal transplantation in the Nordic countries, 1964–1986. Int J Cancer
11. The Activity in the Hospital Care System 1979
[in Danish]. Copenhagen: Danish National Board of Health; 1981.
12. Classification of Diseases
[in Danish]. Copenhagen: Danish National Board of Health; 1976.
13. Classification of Surgical Procedures and Therapies
, 1st ed [in Danish]. Copenhagen: Danish National Board of Health; 1973.
14. Cancer Incidence in Denmark 1996.
Copenhagen: Danish National Board of Health; 1999.
15. ICD O International Classification of Diseases for Oncology
, 1st ed. Geneva: World Health Organization; 1976.
16. Bailar JC, Ederer F. Significance factors for the ratio of a Poisson variable to its expectation. Biometrics
17. Breslow NE, Day NE. Statistical Methods in Cancer Research. Volume II—The Design and Analysis of Cohort Studies.
Lyon: International Agency for Research on Cancer; 1987.
18. Garrel D. Burn scars: a new cause of vitamin D deficiency? Lancet
19. Warden GD. Outpatient care of thermal injuries. Surg Clin North Am
20. Zeitlin RE, Jarnberg J, Somppi EJ, et al. Long-term functional sequelae after paediatric burns. Burns
21. Klein GL, Chen TC, Holick MF, et al. Synthesis of vitamin D in skin after burns. Lancet
22. Frisch M, Biggar RJ, Engels EA, et al, AIDS-Cancer Match Registry Study Group. Association of cancer with AIDS-related immunosuppression in adults. JAMA