Dose Effect in Adjuvant Radiation Therapy for the Treatment of Resected Keloids

Abstract

Purpose:

Surgical excision of keloids can result in an insidious cycle of tissue injury and repeat keloid formation unless combined with adjuvant therapy to halt this cycle. We present our results of postoperative radiation therapy for keloids with various dose regimens.

Methods and Materials:

A retrospective review of 124 patients with 250 keloid lesions treated with postoperative radiation therapy was analyzed. In this institutional review board—approved study, 125 keloids were treated to 20 Gy in 5 fractions and 125 keloids were treated to 12 to 16 Gy in 3 to 4 fractions. Local failure was defined as redevelopment of any clinically apparent keloid at the treated site. The median age was 34 years (14-84 years). Keloids were located on the ear (34%), neck/shoulder (19%), abdomen (13%), chest (10%), face (9%), breast (7%), extremities (4%), and back (3%). Median keloid size was 4 cm (0.5-20 cm).

Results:

At a median follow-up of 40 months, the recurrence rate for all lesions was 5.6%. Lesions treated to 20 Gy had a recurrence rate of 1.6% compared with 9.6% with <20 Gy and an odds ratio of 0.16 (P = .02). Upon univariate and multivariate analysis there were no differences in recurrence rate with respect to location, race, gender, age, previously treated lesions, and presence of multiple keloids. The lone predictor for improved control rate was the dose of 20 Gy in 5 fractions compared with less than that. Control rate for lesions treated to a biologically equivalent dose₂ of 35 to 36 Gy₂, 48 to 52.5 Gy₂, and 60 to 72 Gy₂ were 10% (P = .007), 8.9% (P = .16), and 1.6% (P = .02), respectively.

Conclusions:

Surgical excision followed by immediate adjuvant radiation therapy for keloids provides excellent local control and cosmesis. Treatment with a biologically equivalent dose₂ > 60 (20 Gy in 5 fractions) yielded superior local control over lower dose regimens.

Summary

In the adjuvant treatment of resected keloids, lesions treated to 20 Gy had a recurrence rate of 1.6% compared with 9.6% with <20 Gy (P = .02). Upon univariate and multivariate analysis there were no differences in recurrence rate with respect to location, race, gender, age, previously treated lesions, and presence of multiple keloids. Surgical excision followed by adjuvant radiation therapy for keloids provides excellent local control with a biologically equivalent dose₂ > 60 (20 Gy in 5 fractions).

Introduction

Manifesting as excessive scar tissue extending beyond the initial wound margins, a keloid is a disfiguring and symptomatic condition that affects patients of all Fitzpatrick skin types (1, 2). Pathologically, it has been proposed that keloids progress from aberrant tissue remodeling with disordered fibroblast proliferation resulting from excessive cytokine production and collagen deposition (3, 4). Consequently, patients can experience symptoms of pain, pruritus, and anatomic dysfunction (5). Furthermore, keloids may cause significant cosmetic disfigurement and psychosocial issues, with impaired quality of life (1, 6). Accordingly, patients frequently seek treatment for symptomatic keloids (2).

Despite multiple treatment options, there are no consensus guidelines to optimize management (or no universally accepted standards of care) (7). Intralesional steroids are often the first therapy utilized, with response rates ranging from 50% to 100% and recurrence rates of 9% to 50% (8). Typically, the corticosteroids are given over a series of injections, 2 to 3 over 4 weeks; however, treatment can continue for several months and can be painful (9). Other therapies, such as silicone gel sheets, lack strong evidence of benefit and require use over multiple months, 12 to 24 hours a day. Additional alternative management such as compression therapy is uncomfortable for most patients and requires use for 6 to 12 months (9-11).

Despite conservative treatment, keloids frequently are recalcitrant with high rates of relapse (10). Thus, for many patients surgical excision is performed. Surgical excision alone can result in an insidious cycle of tissue injury and repeat keloid formation with a recurrence rate that ranges from 45% to 100% (12). To halt this cycle, adjuvant radiation therapy can be delivered immediately after surgery and has been shown to be effective at reducing keloid recurrence (13, 14). Various radiation dose and fractionation schema have been employed, and there is no universally accepted dose. There has been a suggestion of a radiation dose-response relationship, with a meta-analysis by Flickinger concluding that moderately high doses are needed for postoperative radiation therapy to be effective (15, 16).

This article reports our experience with keloid treatment at 2 institutions utilizing similar radiation setups but delivering different postoperative radiation doses. We analyzed the clinical outcomes and toxicity associated with the postoperative radiation treatment of keloids, while investigating a dose-response relationship.

Methods and Materials

Patient selection and analysis

Between April 2009 and January 2017, 266 consecutive keloid lesions treated postoperatively with radiation therapy at 2 separate institutions were reviewed in this institutional review board–approved study. In total, 124 patients with 250 keloids had sufficient follow-up data for inclusion in this pooled analysis. All keloid sizes, number of lesions, location, and patient ages were included. Patients were not excluded for prior treatment, including prior irradiation (n = 3) (Table 1). Half of the keloids (n = 125) were treated to 20 Gy in 5 fractions, and the other half were treated to 12 to 16 Gy in 3 to 4 fractions, providing the rationale for a case-control study of outcomes based on dose. The 20 Gy arm and the 12 to 16 Gy arm of this study were delivered in a similar fashion; however, the majority in each arm corresponded with different institutions.

Table 1.

Patient and Keloid Characteristics

Characteristics	Total number	20 Gy group Number (%)	<20 Gy group Number (%)	P value
Patient Characteristics
Black	213	112 (90%)	101 (81%)	.053
White	32	10 (8%)	22 (18%)	<.01
Male	64	26 (21%)	38 (30%)	.08
Female	186	99 (79%)	87 (70%)	.08
Age	34 (14-84)	34	34	.99
Keloid characteristics
Ear	86	37 (30%)	49 (39%)	.11
Head and neck	70	41 (33%)	39 (31%)	.78
Torso	83	43 (34%)	40 (32%)	.68
Extremity	11	4 (3%)	7 (6%)	.36
Multiple lesions	70	30 (24%)	40 (32%)	.16
Prior surgery	40	9 (7%)	31 (25%)	<.01
Prior steroid injection	48	17 (14%)	21 (17%)	.48

Treatment

All lesions underwent local excision followed by postoperative radiation therapy within 48 hours of surgery (≥90% within 24 hours). The radiation technique was uniform for all treatments: linear accelerator-based external beam radiation was delivered enface with 6 MeV electrons to the surgical incision, prescribed to 90% depth, and typically with a 5 mm bolus. Custom Cerrobend blocks were applied on the basis of clinically drawn block margins of 1.5 to 2 cm radially from the incision. Patients were treated daily (except for weekends) to between 12 and 20 Gy in 3 to 5 fractions, with a corresponding biologically equivalent dose₂ (BED₂) ranging from 36 Gy₂ to 72 Gy₂ (median 60 Gy₂). An α/β ratio of 2 was selected for dose comparison on the basis of modeling from a large meta-analysis, which indicated that keloids behave as late-reacting tissue (15). Patients with multiple lesions were treated concurrently. The prescribed dose was arbitrary and not based on location, keloid history, or size of the lesion/surgical incision. Fifty-six lesions were treated previously with surgery (n = 8), steroid injection (n = 16), or a combination of both (n = 32).

Outcome assessment

The median follow-up of this study was 40 months (range 3-146 months). The primary objective was to evaluate the recurrence rate and its correlation with radiation dose, particularly for a prescription dose of 20 Gy in 5 fractions compared with a lower total dose. Secondarily, we evaluated the impact of age, race, site, number of lesions, and prior therapy on local control. Four lesions treated to 18 Gy in 3 fractions were included in the 20 Gy in 5 fraction group given the higher BED. A local recurrence was defined as any clinical evidence of a keloid developing at the incision site, regardless of the size, as determined with a focused physical examination by the treating radiation oncologist, dermatologist, plastic surgeon, or otolaryngologist in follow-up. Student t-tests and chi squares were used to determine statistical similarities between the 20 Gy and <20 Gy cohorts, and both univariate and multivariate analyses were performed via binary logistic regression models on SPSS version 20 (IBM SPSS, Inc).

Results

Patient characteristics

Eighty-seven females and 37 males with 250 keloids were treated postoperatively to either 20 Gy in 5 fractions (n = 121), 18 Gy in 3 fractions (n = 4), 16 Gy in 4 fractions (n = 31), 15 Gy in 3 fractions (n = 14), 12 Gy in 3 fractions (n = 78), or 10 Gy in 2 fractions (n = 2) (Table 2). The median age of all patients was 34 years (range 14-84 years). Multiple lesions were treated in 70 patients, and 60 lesions were recurrent, having been treated previously with surgery, steroid injection, or a combination of both. In order of most to least common sites: 86 (34%) keloids were located on the ear, 48 (19%) on the neck/shoulder, 32 (13%) on the abdomen, 26 (10%) on the chest, 22 (9%) on the face, 17 (7%) on the breast, 11 (4%) on the extremities, and 8 (3%) on the back. Incision sizes were not recorded for all patients, but for the available data sizes ranged from 0.5 to 20 cm (median 4 cm). There were disproportionately more white patients, previously treated lesions, and previous surgeries in the <20 Gy group, and a trend toward disproportionately more black and female patients in the 20 Gy group (Table 1).

Table 2.

Univariate and Multivariate Analysis for Keloid Recurrence

Variable	N (%)	Recurrence rate	P-value (Univariate)	P-value (Multivariate)
Dose
20 Gy in 5 Fx	125 (50%)	1.6%	.015	.015
<20 Gy in 5 Fx	125 (50%)	9.6%
Site
Ear	86 (57%)	7.0%	.523	.868
Head and neck	70 (28%)	4.3%	.575
Torso	83 (33%)	6.0%	.961
Extremity	11 (4%)	0%	.828
Prior therapy
Prior surgery	40 (16%)	7.5%	.667	.423
No prior surgery	210 (84%)	5.2%
Prior steroid injection	48 (19%)	8.3%	.414	.585
No prior steroid injection	202 (81%)	5.0%
Patient characteristics
Black	213 (85%)	6.1%	.420	.999
White	32 (13%)	3.1%
Male	64 (26%)	7.8%	.377	.496
Female	186 (74%)	4.8%
34 years old or older	137 (55%)	7.3%	.208	.452
Under 34 years old	113 (45%)	3.6%

Outcomes

At a median follow-up of 40 months, the overall recurrence rate for all lesions was 5.6%. Lesions treated in the 20 Gy cohort had a recurrence rate of 1.6% compared with 9.6% in the <20 [Gy cohort with an odds ratio of 0.16 (confidence interval [CI] 0.036-0.75, P = .02). Upon univariate and multivariate analysis there were no statistically significant differences between recurrence rate and location, race, gender, age, previously treated lesions, and presence of multiple keloids (Table 2). The lone predictor for improved control rate with both univariate and multivariate analysis was the dose of 20 Gy in 5 fractions. A dose response was demonstrated as the control rate for lesions treated to a BED₂ of 35 to 36 Gy₂, 48 to 52.5 Gy₂, and 60 to 72 Gy₂; these were 10% ( P= .007), 8.9% (P = .16), and 1.6% (P = .02), respectively. The primary corresponding doses for low, intermediate, and high BED₂ groups were 12 Gy in 3 fractions, 16 Gy in 4 fractions, and 20 Gy in 5 fractions, respectively. A dose-response chart is depicted in Figure 1.

Fig. 1.

Recurrences by dose group. Abbreviations: BED = biologically equivalent dose; Fx = fraction; Gy = Gray.

Toxicity

There were no instances of acute or late grade 3 toxicity as determined by the Common Terminology for Criteria for Adverse Events version 4.0 (17). There were also no reported second malignancies at the treatment sites. One patient who was treated to 4 separate areas of the face developed breast cancer, and another patient treated at the ear developed lung cancer several years after treatment. Grade 2 neuropathic pain developed in the right axillary region in a 78-year-old woman who had keloids in the right and left axilla and right upper breast, all of which were treated to 12 Gy in 3 fractions. A 42-year-old woman whose abdomen was treated using 15 Gy in 3 fractions had wound healing difficulties following surgery and radiation.

Discussion

There is no universally accepted radiation dose in the postoperative management of resected keloids, and the literature reports numerous dose and fractionation schedules (15). In 2015 the Deutsche Gesellschaft fur Radioonkologie (DEGRO) German Cooperative Group on Radiotherapy of Benign Diseases (GCG-BD) recommended postkeloidectomy radiation therapy to a total dose of 16 to 20 Gy in 5 fractions over 1 week delivered within 24 hours of resection. They stated that this would result in an estimated recurrence rate of 10% to 30% (18). To our knowledge, this is the only cooperative group consensus on a potential dose in the adjuvant treatment of keloids. Our study offered us the unique opportunity to evaluate this consensus dose versus a more conservative low dose between keloids treated in a similar fashion at 2 institutions.

The effectiveness of different dose and fractionation schedules is confounded by different definitions of recurrence, duration of follow-up, and heterogeneity of patient population and keloid characteristics, including the disproportionately higher number of white patients and previously excised lesions in the high-dose group. Accordingly, comparison of effectiveness between different reports is difficult and makes establishing a standardized dose and fractionation challenging. In our study we sought to simplify these confounders. Therefore, we defined recurrence as any redevelopment of clinically apparent keloid to simplify a meaningful endpoint. With this definition, our overall recurrence rate in 250 treated keloids was 5.6%. Additionally, median time to development of recurrence has been reported to be 12 to 15 months, and sufficient follow-up is necessary to evaluate recurrence (19-21). In our study the median follow-up was 40 months, which is sufficient time for evaluation. This study was homogeneous and well balanced with proportionate risk factors shown to be associated with keloid recurrence such as sex, race, high tension sites, keloid size, history of previous treatments, and time to start of radiation therapy (13, 14, 19, 20).

Various reports suggest a dose and time response in the treatment of keloids. Jiang et al performed an in vitro analysis of fibroblast cultured from de novo resected keloid specimens and subjected them to radiation therapy with differing dose and time effects (15, 16, 22). They showed a dose and time dependent induction of keloid fibroblast senescence and postulated that this cell cycle arrest is the mechanism behind improved control after postoperative radiation therapy for resected keloids. Improved control rates when keloids are treated within 24 to 48 hours has been reported numerous times and is widely accepted (20, 23). A dose effect is less established but has been shown clinically and is detailed in 2 large meta-analyses (15, 16).

Kal et al performed a meta-analysis in 2005 and showed with a BED₁₀ >30 Gy that the recurrence rate was <10% (16). This corresponded to conventional doses of 16 Gy (8 Gy × 2) and 18 Gy (6 Gy × 3) (24). The meta-analysis was subsequently confirmed with a retrospective review of 826 keloids treated with 18 Gy in 2 treatment fractions 1 week apart. Shen et al showed a recurrence rate of 9.59% at the expense of a 9.83% risk of chronic late adverse effects including skin hyperpigmentation and/or telangiectasia with depigmentation with this moderately high dose per fraction (20). Flickinger subsequently published a meta-analysis in 2011 indicating that using an α/β ratio of 2.08 the 3 fraction dose to achieve 90% local control was 16.0 to 16.2 Gy for the ear lobe and 21.5 to 22.2 Gy for nonearlobe sites (15). These studies suggest a dose-response effect; however, owing to variable radiation techniques (eg, electron beam, orthovoltage photons, brachytherapy) included in the Flickinger meta-analysis, confirmation that a dose of approximately 20 Gy in 5 fractions results in 90% control without added toxicity has not been established compared with a low dose arm (15).

We took advantage of consistently different dose schedules between 2 institutions treating in a standard fashion with electron beam radiation therapy to analyze for dose response. In our study lesions treated in the 20 Gy cohort had a recurrence rate of 1.6% compared with 9.6% in the <20 Gy cohort with an odds ratio of 0.16 (CI 0.036-0.75, P = .02). A dose response was demonstrated as the control rate for lesions treated to a BED₂ of 35 to 36 Gy₂, 48 to 52.5 Gy₂, and 60 to 72 Gy₂ were 10% (P = .007), 8.9% (P = .16), and 1.6% (P = .02), respectively. The primary corresponding doses for low, intermediate, and high BED₂ groups were 12 Gy in 3 fractions, 16 Gy in 4 fractions, and 20 Gy in 5 fractions, respectively. These results support the findings by Kal et al and Flickinger and indicate that a BED₂ > 35 Gy yields favorable local control across all keloid sites (15, 16).

Acute and chronic toxicity was infrequent in our study. We postulate that the moderately hypofractionated doses used in our study (4-6 Gy/fraction) resulted in effective control without added late toxicity as seen in the 9 Gy/fraction review by Shen et al (20). Furthermore, in our patients with 40 months median follow-up there have been no instances of second malignancies attributable to radiation therapy as has been shown in multiple other reports (20-22, 25). However, it should be noted that our follow-up time is short compared with the latency period for development of radiation-induced malignancy.

This study is limited by the inherent selection bias of retrospective review, partially exemplified by the higher proportion of previously treated lesions in the <20 Gy group. Nevertheless, total dose was the sole predictor of recurrence in this study by a considerable margin, and the influence of confounding variables was mitigated with multivariate analysis. Treatment technique was largely uniform with the exception of block margin, for which the variation was minimal.

Conclusion

Postoperative radiation therapy for treatment of resected keloids yields excellent results. In this series a dose-response effect on recurrence rate was demonstrated. While prospective validation is warranted, a dose of 20 Gy in 5 fractions may be superior to 12 to 16 Gy in 3 to 4 fractions in the treatment of resected keloids across all sites.