Thomson et al.1 searched the Cochrane Skin Group Specialized Register, the Cochrane Central Register of Controlled Trials, MEDLINE and Embase via Ovid, CINAHL via EBSCO, LILACS, and 5 clinical trial registers. They also checked the references of the studies included, and where necessary contacted authors for additional information. The date of the last search was November, 2019. Three authors independently reviewed the titles, abstracts, and full texts, and discrepancies were resolved by a fourth author.

The authors included randomized controlled trials (RCTs) evaluating surgical and nonsurgical interventions for the treatment of any type of BCC in immunocompetent patients with a biopsy-confirmed diagnosis. They excluded studies of patients with persistent or recurrent tumors and syndromes with a high risk of BCC (e.g., Gorlin syndrome).

Comparators were placebo, active treatment, other treatments, or no treatment.

The primary outcome measures were 1) clinical recurrence at 3–5 years or any time if no information was available for this period and 2) cosmetic outcomes rated as good or excellent by the patient or an observer (investigator).

Secondary outcome measures were 1) pain during and after treatment, 2) treatment failure in the first 6 months, and 3) adverse events (AEs).

Bias was assessed using the Cochrane risk of bias tool. Risk was assessed separately by 2 authors, and discrepancies were resolved by discussion with a third.

Quality of evidence for each outcome was assessed using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) framework.

Thirty-two trials had a low risk of bias for incomplete outcome data, while 19 had an unclear risk. One trial had a high risk of bias in this area due to a large between-group difference in the number of participants who withdrew.

Just 11 trials preregistered the study protocol, and of these, 6 had a low risk of bias for selective reporting of outcomes. Forty-four trials had an unclear risk of bias in this area, while 2 had a high risk.

Eight trials were described as pilot studies. Twenty trials compared one nonsurgical treatment with another, while 14 trials compared a nonsurgical treatment with placebo. Surgical treatment was used as a comparator in 18 trials. It was compared with a nonsurgical treatment in 10 trials, another surgical treatment in 5, and placebo in 3.

Just 1 trial, involving 374 participants and 408 primary BCCs with a high-risk histologic subtype located in the H zone of the face, compared MMS and SE.

The authors detected slightly fewer recurrences for MMS compared with SE at both 3 years (1.9% vs. 2.9%; risk ratio [RR], 0.64; 95% CI, 0.16–2.64; low-certainty evidence) and 5 years (3.2% vs. 5.2%; RR, 0.61; 95% CI, 0.18–2.04; low-certainty evidence) (Fig. 1A, B).

Figure 1.

(1.1 and 1.2) Mohs micrographic surgery (MMS) vs. surgical excision (SE). Outcome: recurrence at 3 and 5 years (analyses 1.1 and 1.2 in the original review1). Interpretation: moderate-certainty evidence. Just 1 study analyzed recurrence at 3 and 5 years. Evidence in favor of SE for reduction in risk of recurrence at 3 and 5 years, but low certainty due to imprecision (wide confidence interval). MH indicates Mantel–Haenszel.

(0.11MB).

The evidence supporting the effects of MMS and SE on cosmetic outcomes and adverse events (AEs) is uncertain as these were not analyzed in the original trial.

One noninferiority trial compared imiquimod with SE in 501 patients with nodular and superficial BCCs in low-risk areas. The authors of the Cochrane review concluded that imiquimod probably results in higher recurrence rates than SE at both 3 years (16.4% vs. 1.6%; RR, 10.30; 95% CI, 3.22–32.94; moderate-certainty evidence) and 5 years (17.5% vs. 2.3%; RR, 7.73; 95% CI, 2.81–21.30; moderate-certainty evidence) (Fig. 2A, B).

Figure 2.

(2.1 and 2.2) Imiquimod vs. surgical excision (SE). Outcome: recurrence at 3 and 5 years, respectively (analyses 2.1 and 2.2 in the original review1). Interpretation: moderate-certainty evidence. Just 1 study analyzed recurrence at 3 and 5 years. Evidence in favor of SE for reduction in risk of recurrence at 3 and 5 years with a wide confidence interval, indicating imprecision. (2.3) Imiquimod vs. SE. Outcome: good/excellent cosmetic outcome (analysis 2.3 in the original review1). Interpretation: low-certaintly evidence in favor of imiquimod. Just 1 study with a risk of blinding bias. (2.4) Imiquimod vs. SE. Outcome: moderate/severe pain (analysis 2.4 in the original review1). Interpretation: low-certainty evidence in favor of imiquimod during follow-up. Just 1 study with a high risk of blinding bias and dropout bias in the evaluation of pain during treatment in the SE group. MH indicates Mantel–Haenszel.

(0.29MB).

No differences were observed between the 2 treatments for patient-rated cosmetic outcomes at 6 months or 3 years, with 91.6% of patients in the imiquimod group and 92.2% of those in the SE group reporting good or excellent outcomes (RR, 1.00; 95% CI, 0.94–1.06; low-certainty evidence). Observer-rated outcomes, by contrast, were better for imiquimod, with 60.6% deemed good or excellent compared with 35.6% for SE (RR, 1.70; 95% CI, 1.35–2.15; low-certainty evidence) (Fig. 2C).

A higher proportion of patients treated with imiquimod reported pain during treatment (30% vs. 22% for those who underwent SE; RR, 1.36; 95% CI, 0.98–1.88; low-certainty evidence). The opposite, however, was observed, at 16 weeks, with just 9% of patients in the imiquimod group reporting pain compared with 20% in the SE group (RR, 0.47; 95% CI, 0.29–0.77; low-certainty evidence) (Fig. 2D).

Mild to moderate AEs were slightly more common in the imiquimod group (94% vs. 88% in the SE group). Itching and weeping were considerably more common in patients treated with imiquimod compared with SE (85% vs. 56% and 64% vs. 35%, respectively). In the only trial to analyze AEs, 5 patients in the imiquimod group (vs. none in the SE group) dropped out due to treatment-related AEs. Thirty-eight patients treated with imiquimod (15%) required a dose reduction due to AEs.

Just 1 trial compared radiotherapy and SE with or without frozen section margin control in 374 patients with facial BCC (low- and high-risk histologic subtypes) measuring less than 4cm. Radiotherapy was associated with a recurrence rate of 5.2% at 3 years compared with 0% for SE (RR, 19.11; 95% CI, 1.12–325.78; low-certainty evidence). Recurrences were also more common with radiotherapy at 4 years (6.4% vs. 0.6% for SE) (RR, 11.06; 95% CI, 1.44–84.77; low-certainty evidence) (Fig. 3A, B).

Figure 3.

(3.1 and 3.2) Radiotherapy vs. surgical excision (SE) with and without frozen section margin control. Outcome: recurrence at 3 and 4 years (analyses 3.1 and 3.2 in the original review1). Interpretation: low-certainty evidence in favor of SE based on just 1 study with a risk of bias due to indirect evidence. Imprecision (wide confidence intervals) in the evaluation of the effect on recurrence at 3 and 4 years. (3.3) Radiotherapy vs. SE with and without frozen section margin control. Outcome: patient- and observer-rated cosmetic outcome (analysis 3.3 in the original review1). Interpretation: moderate-certainty evidence in favor of SE from just 1 study with a risk of blinding bias. Narrow confidence intervals for estimated effect on patient- and observer-rated cosmetic outcomes. MH indicates Mantel–Haenszel.

(0.26MB).

Cosmetic outcomes assessed on a clinical scale of poor, fair, and good at 4 years were more likely to be rated unfavorably in the radiotherapy group compared with the SE group (RR, 0.76; 95% CI 0.63–0.91; moderate-certainty evidence) (Fig. 3C).

Similar results were observed for observer-rated outcomes (RR 0.48; 95% CI, 0.37–0.62; moderate-certainty evidence).

Dyspigmentation and telangiectasia at 4 years were more common in patients treated with radiotherapy (>65%), as were cutaneous radiodystrophy (41.5%), necrosis (25%), and scar deformations and constrictions (5%).

SE was compared with MAL-PDT in 3 trials12–14 and ALA-PDT in 1.15 The trial by Rhodes et al.12 included 103 patients with a total of 118 nodular facial BCCs and follow-up of 5 years. The authors, however, performed a per-protocol analysis, providing insufficient data for Thomson et al.1 to determine the denominators for recurrence rates. Compared with SE, MAL-PDT appears to be associated with better patient- and observer-rated cosmetic outcomes (RR, 1.17; 95% CI, 1.01–1.34 and RR, 2.08; 95% CI, 1.38–3.12, respectively), but more pain and a burning sensation of the skin (13.4% vs. 6.1%; RR, 2.20; 95% CI, 0.60–8.03). Overall, AEs (burning sensation of the skin, pain in the skin, and erythema) were more probable in patients treated with MAL-PDT (52% vs. 29%) (P=.03, Fisher exact test). Three patients in the SE group developed a severe skin infection. Abbade et al.14 found that MAL-PDT was more likely to result in recurrence than SE (36.36% vs. 0%; RR, 26.47; 95% CI, 1.63–429.92) in a study of 57 patients with nodular BCC in the head and neck area (68 lesions) and a follow-up of more than 3 years. The risk of early treatment failure was also higher for PDT (RR, 11.65; 95% CI, 0.67–202.74)1,14 (Fig. 4A).

Figure 4.

(4.1) Photodynamic therapy with methylaminolevulinate (MAL-PDT) vs. surgical excision (SE). Outcome: recurrence at 3 years (analysis 10.1 in the original review1). Interpretation: low-certainty evidence from just 1 study with a high risk of bias and imprecision in the evaluation on the estimated effect of recurrence at 3 years (wide confidence interval [CI]). (4.2) MAL-PDT vs. SE. Outcome: good/excellent cosmetic outcome (analysis 10.2 in the original review1). Interpretation: moderate-certainty evidence in favor of MAL-PDT from a single study with a high risk of blinding bias, with narrow CIs for patient-rated outcomes and wide CIs (imprecision) for observer-rated outcomes. MH indicates Mantel–Haenszel.

(0.19MB).

In a noninferiority trial, Szeimies et al.13 compared MAL-PDT and SE in 196 patients with 246 superficial BCCs located anywhere except the facial H zone; they were followed for 1 year.13 The inferiority margin was a difference of 15% in the percent reduction in lesion count 3 months after the last treatment. Both patient- and observer-rated cosmetic outcomes were more favorable for MAL-PDT, with respective RRs of 1.18 (95% CI, 1.08–1.30) and 1.81 (95% CI, 1.46–2.25) (Fig. 4B).

Szeimies et al.13 reported pain during and after treatment, attributing it to photosensitivity reactions that are expected to occur with PDT (e.g., skin discomfort, burning sensation, stinging, and erythema). AEs were reported for 37% of patients treated with MAL-PDT vs. 14% of those who underwent SE. One patient in the SE group developed a severe skin infection.

In a study of 149 patients with 171 nodular BCCs and a follow-up of 5 years, Mosterd et al.15 observed a 3-year recurrence rate of 24.7% for those treated with ALA-PDT and 2.3% for those treated with SE (RR, 10.87; 95% CI, 2.63–44.95; moderate-certainty evidence). ALA-PDT also appeared to result in more recurrences at 5 years (27.1% vs. 2.3%; RR, 11.91; 95% CI, 2.90–48.95; moderate-certainty evidence). SE appeared to improve the risk of early treatment failure compared with ALA-PDT (7.2% vs. 2.3%; RR, 3.18; 95% CI, 0.66–15.32; low-certainty evidence). One severe infection was also reported following SE in this trial.

Arits et al.1,16 conducted a noninferiority trial with a follow-up time of 5 years comparing 5-fluorouracil (5-FU) cream and imiquimod cream with PDT-MAL in 601 patients with a single high-risk BCC located anywhere except the face or scalp. Imiquimod compared with MAL-PDT probably reduces the risk of recurrence at both 3 years (22.8% vs. 51.6%; RR, 0.44; 95% CI, 0.32–0.62; moderate-certainty evidence) and 5 years (28.6% vs. 68.6%; RR, 0.42; 95% CI, 0.31–0.57; moderate-certainty evidence) (Fig. 5A, B).

Figure 5.

(5.1 and 5.2) Imiquimod vs. photodynamic therapy with methylaminolevulinate (MAL-PDT). Outcome: recurrence at 3 and 5 years, respectively (analyses 18.1 and 18.2 in the original review1). Interpretation: moderate evidence from just 1 study showing favorable results for imiquimod in the reduction of recurrence at 3 and 5 years. MH indicates Mantel–Haenszel.

(0.14MB).

No significant differences were observed for good or excellent cosmetic outcomes at 1 year (RR, 0.98; 95% CI, 0.84–1.16; moderate-certainty evidence).1,16

Thomson et al.1 showed that imiquimod probably reduces the risk of severe pain compared with MAL-PDT (RR, 0.60; 95% CI, 0.41–0.87; moderate-certainty evidence).

Nonsignificant differences were observed for treatment failures at 3 months (10.1% for imiquimod vs. 15.8% for MAL-PDT; RR, 0.64; 95% CI, 0.37–1.09; moderate-certainty evidence).

Patients treated with imiquimod had higher rates of edema, erosion, crust formation, and itching; 4.8% developed a serious AE (8 cases of flu-like symptoms and 1 local wound infection treated in an outpatient setting).

In the trial by Arits et al.16 imiquimod seemed to reduce the risk of recurrence at 3 years compared with topical 5-FU (23.4% vs. 34.2%; RR, 0.68; 95% CI, 0.47–0.99; moderate-certainty evidence). A similar finding was made for recurrence at 5 years (28.6% vs. 46%; RR, 0.62; 95% CI, 0.44–0.87; moderate-certainty evidence)1,16 (Fig. 6A, B).

Figure 6.

(6.1 and 6.2) Imiquimod vs. 5-fluorouracil (5-FU). Outcome: recurrence at 3 and 5 years, respectively (analyses 23.1 and 23.2 in the original review1). Interpretation: moderate evidence from just 1 study showing favorable results for imiquimod in the reduction of recurrence at 3 and 5 years. Inconclusive findings. (6.3) Imiquimod vs. 5-FU. Outcome: good/excellent observer-rated cosmetic outcome at 1 year (analysis 23.3 in the original review1). Interpretation: Moderate evidence with just 1 study in favor of imiquimod for cosmetic outcome. Inconclusive findings. (6.4) Imiquimod vs. 5-FU. Outcome: moderate/severe pain (analysis 2.4 in the original review1). Interpretation: Moderate evidence with just 1 study in favor of 5-FU for cosmetic outcome. Inconclusive findings. (6.5) Imiquimod vs. 5-FU. Outcome: early treatment failure (analysis 23.5 in the original review1). Interpretation: low evidence with just 1 study in favor of imiquimod for early treatment failure. Inconclusive findings. MH indicates Mantel–Haenszel.

(0.37MB).

No significant differences were observed for cosmetic outcomes rated as good or excellent at 1 year (61.4% for imiquimod vs. 57.5% for 5-FU; RR, 1.07; 95% CI, 0.90–1.26; moderate-certainty evidence) (Fig. 6C).

Imiquimod was associated with a slightly higher risk of pain (18.2 vs. 12.5%; RR, 1.46; 95% CI, 0.89–2.34; moderate-certainty evidence) (Fig. 6D).

No significant differences were observed for early treatment failures (10% for imiquimod vs. 12.1% for 5-FU; RR, 0.83; 95% CI, 0.47–1.46; low-certainty evidence) (Fig. 6E).

Patients treated with imiquimod and 5-FU experienced higher rates of edema, erosion, crusting, and pruritus than those treated with MAL-PDT. Two patients in the 5-FU group developed a local wound infection managed in an outpatient setting. This treatment was also associated with erysipelas of the lower limb in one patient and a leg ulcer in another.

On comparing MAL-PDT and cryotherapy in 118 participants with 219 BCCs followed for 5 years, Basset-Seguin et al.17 found no significant differences in recurrence at 3 years (22% vs. 19.4%; RR, 1.14; 95% CI, 0.65–1.98; low-certainty evidence) or 5 years (22% vs. 20%; RR, 1.08; 95% CI, 0.62–1.86; low-certainty evidence).1 The treatments also resulted in similar rates of pain (33% for MAL-PDT vs. 37% for cryotherapy; RR, 1.12; 95% CI, 0.68–1.84; low-certainty evidence)1,17 (Fig. 7A, B).

Figure 7.

(7.1 and 7.2) Photodynamic therapy (PDT) vs. cryosurgery. Outcome: recurrence at 3 and 5 years, respectively (analyses 16.1 and 16.2 in the original review1). Interpretation: low evidence with just 1 study with a high risk of reporting bias for recurrence at 3 and 5 years. Inconclusive findings. (7.3) PDT vs. cryosurgery. Outcome: patient- and observer-rated cosmetic outcome (analysis 16.3 in the original review1). Interpretation: low-certainty evidence from just 1 study with a high risk of reporting bias for patient- and observer-rated cosmetic outcomes. The findings suggest that PDT results in better cosmetic outcomes than cryotherapy. (7.4) PDT vs. cryosurgery. Outcome: pain (analysis 16.4 in the original review1). Interpretation: low-certainty evidence from just 1 study with a high risk of reporting bias for pain. Confidence interval crossing 1, indicating inconclusive evidence on whether PDT causes more or less pain. In other words, the evidence is inconclusive.

(0.29MB).

MAL-PDT appears to result in higher rates of cosmetic outcomes rated as good or excellent by patients at 1 year (100% vs. 81.3% for cryotherapy; RR, 1.23; 95% CI, 1.07–1.41; moderate-certainty evidence).1,17 Similar findings were noted for observer-rated outcomes at 1 year, with 100% of MAL-PDT outcomes rated as good or excellent at 1 year compared with 61% of cryotherapy outcomes (RR, 1.46; 95% CI, 1.07–1.88, moderate-certainty evidence) (Fig. 7C).1,17 Wang et al.18 compared ALA-PDT with cryotherapy in 88 patients with nodular and superficial BCC. Risk of recurrence was not analyzed in the Cochrane review1 as the follow-up time was just 1 year. ALA-PDT also appears to be associated with an increased likelihood of good or excellent cosmetic outcomes compared with cryotherapy at 1 year (92.8% vs. 54.1%; RR, 1.72; 95% CI, 1.26–2.34; moderate-certainty evidence).

ALA-PDT is probably more painful than cryotherapy (difference in means, 11.00; 95% CI, 1.12–23.12; moderate-certainty evidence). In the study by Wang et al.,18 2 patients died of a cause unrelated to BCC or treatment. One patient treated with ALA-PDT described pain radiating from the treatment site, while another, treated with cryotherapy, developed a bacterial infection.1,18 (Fig. 7D).