We conducted initial title and abstract screening. Publications selected for eligibility assessment were evaluated by full-text reading by two authors (J.V.G. and S.G.). Disagreements were resolved by a third member (M.A.).

Risk of bias (ROB) was assessed according to the recommendation by the Cochrane Collaboration, and was independently evaluated by two reviewers.19 Case reports were assumed to present multiple bias.20 The following items were assessed: selection biases, performance biases, detection biases, attrition biases, reporting biases, and other sources of bias (e.g., potential conflicts of interests, modifications of conditions influencing measurements). These items were classified as “low risk” (when no bias was identified), “high risk” (when bias was detected), or “unclear risk” (when there was not enough information). Quality rating of studies was done as follows: 1: RCT and systematic review with meta-analysis; 2: non-randomized controlled trials or prospective comparative cohort trials; 3: case–control studies and retrospective cohort studies; 4: case series and cross-sectional studies; and 5: case reports and opinion of experts.

The percentage of studies with each risk grade are represented in Fig. 2, based on the researchers’ evaluation (see Supplementary data 6 for detailed information).

Fig. 2.

Summary of review authors’ judgements for each risk of bias domain. The graph shows the proportion of articles presenting a “Low risk”, “High risk” or “Unclear risk” of bias for each item.

Forty-one articles studying SCC and other squamous premalignant conditions were included (34 RCTs, large observational studies or case series, and 7 case reports), including AKs (31/41), actinic cheilitis (2/41), SCC (including Bowen's disease) (7/41), and Bowenoid papulosis (1/41) (Table 1 and Supplementary Tables 1 and 3). Regarding treatment modalities, there was only 1 study21 assessing the effectiveness of ablative laser for the treatment of AKs in 30 patients, showing a reduction of RCM features suggesting subclinical AKs. Two articles15,22 studied the efficacy profile of cryosurgery for treating AKs, reporting an improvement of the main features by RCM. Seven articles2,3,8,14,23–25 studied the efficacy profile of daily 5-FU 0.5% for 2–12 weeks in AKs, showing a complete response rate between 33.3% and 100%. The efficacy profile of IMQ 3.75–5% was assessed in 6 studies,4,13,17,26–28 achieving a complete response in a single case of SCC, and in a 23.1–90% for AKs. Ingenol mebutate 0.015–0.05% was evaluated in two studies, showing a complete response in three cases of actinic cheilitis,29 and an improvement of histological and RCM features right after ending the therapy in 10 cases of Aks.30 Several studies evaluated the efficacy profile of PDT, 3 of them for the treatment of in situ SCCs (13 lesions, including one Bowen's disease case),16,31,32 1 for treating invasive SCC (comprising 4 cases),33 and 13 for treating AKs with conventional or daylight PDT.6,13,15,30,34–42 These works showed a complete response rate of 63.6–100% for in situ SCC, 75% for invasive SCC, and 51.6–100% for AKs. Other therapies such as piroxicam, diclofenac, tirbanibulin, or 2,4,6-octatrienoic acid were used for the treatment of AKs in 5 studies,2,30,43–45 revealing an improvement of RCM features suggesting persistence of AKs. There were also seven reports39,46–50 of different combinations of the abovementioned therapies, showing a complete response rate of 67.7–82.9% for AKs, and 100% for case series of SCC and actinic cheilitis.

Overall, RCM showed restoration of a normal honeycomb pattern and reduction of scaling and parakeratosis (nucleated keratinocytes in the stratum corneum) in patients with complete response. Clinical and dermoscopic evaluation frequently overestimated complete response rate vs RCM, as the latter detected persistence of AKs despite having no clinical lesions. Within the first weeks after treatment, apart from the findings related to the malignant/premalignant condition, inflammation in form of exocytosis, vesicles, and dendritic cells in the epidermis could be detected in RCM, even when clinical and dermoscopic evaluation was not feasible due to local inflammatory reaction (even though intense inflammation could hamper RCM assessment). Inflammatory cells in dermis could also be observed.

When available, histology agreed in most of the cases with findings in RCM.

A total of 15 studies on non-invasive treatment of BCCs were included, including superficial, nodular, and mixed type BCCs (Table 2 and Supplementary Tables 2 and 4). Four articles51–54 studied the efficacy profile of ablative laser (CO2 and Er:YAG), yielding a complete response rate of 33.3–100%, being the response rates higher in superficial BCCs vs the rest of histological subtypes. Two studies47,55 assessed the efficacy profile of cryosurgery, showing a complete response rate of 20–100%. Four studies5,9,56,57 performed treatment with imiquimod 5%, showing a complete response rate of 53.8–100%. Three articles1,7,58 studied PDT with MAL and 5-ALA, presenting a complete response rate of 25–100%. Furthermore, there was 1 study59 in which combination of ablative laser and ingenol mebutate was tested, showing a complete response rate of 61.5–85.7%, and a case report in which ablative laser and conventional PDT were combined, showing a complete response.

In patients with complete response, RCM evidenced disappearance of tumor islands with streaming of basal cells and palisading. Conversely, residual tumor was mainly detected as tumor islands composed of cells with elongated nuclei, accompanied by large blood vessels on some occasions. Overall RCM and histology yielded similar results regarding complete response and treatment failure rates, when both performed; nonetheless, in some studies RCM underestimated the efficacy profile of the therapy, showing findings suggestive of treatment failure, while histology showed a complete clearance.59

One of the included articles60 studied the utility of RCM for monitorization of extramammary Paget's disease treated with imiquimod 5%, 5 times/week for 12 weeks (n=5), and 1 case treated with conventional PDT. A complete response rate of 80% (4/5) in the IMQ group, and 100% (1/1) in the PDT group was observed. One of the patients treated with IMQ presented a recurrence of the disease (time to recurrence not specified).

Despite the scarce literature studying it, CO2 ablative laser proved effective treating clinical evident AK, and even subclinical Aks.21

Regarding cryosurgery, Mota et al.22 compared the effectiveness of 1 vs 2 cycles of cryosurgery, and even though clinical results were similar, RCM proved more effective with 2 cryosurgery cycles. On the other hand, Curiel-Lewandrowski et al.15 concluded that Grade 1 AKs showed a more evident and durable response compared to Grade 2 AKs, and that even though some features may show a reduction at 3 months post-treatment, they could reappear at 6 months. This fact highlights the importance of performing follow-up with tools capable of detecting early recurrences of lesions. Moreover, they also suggested that the most reliable features for monitoring AK response are stratum corneum disruptions, hyperkeratosis, atypical honeycomb pattern including epidermal disarrangement.15

Regarding 5-FU, the available studies detected a reduction in the distinctive RCM findings indicative of AK at the end of the follow-up.2,3,8,14,23–25 Nonetheless, RCM allowed the detection of persistence of subclinical AKs after treatment.3 A RCT compared 5-FU against sodium diclofenac yielding no differences regarding the reduction in the number of AKs or improving field cancerization.2 Ulrich et al.25 compared 5-FU against placebo and despite not finding statistically significant differences in terms of complete response rate for AKs, they did find statistical significance in reducing the number of subclinical AKs. Regarding imiquimod, RCM allowed to detect persistence of AKs even when clinically was free of disease, and thus to guide continuation of the treatment until complete response was achieved.17,26 Furthermore, Benati et al.27 stated that RCM allowed assessment of treated field right 2 weeks after initiation of imiquimod, even though clinical evaluation was not possible due to the inflammatory response. On the other hand, Campione et al.30 studied the utility of ingenol mebutate for AKs, showing that reduction of hyperkeratosis and parakeratosis were the first signs of response, followed by the normalization of the honeycomb pattern.

Regarding PDT, in several studies,34–36,39 clinical and dermoscopic assessment overestimated AK clearance, and thus RCM allowed to administer further PDT cycles when needed. Moreover, Xiang et al.33 suggested that the restoration of a typical honeycomb pattern was the most reliable marker of response to treatment in invasive SCC. Similarly, Curiel-Lewandrowski et al.15 suggested that the improvement of the atypical honeycomb pattern, hyperkeratosis, epidermal disarrangement, and stratum corneum disruptions were reliable markers of treatment response. Moreover, they manifested that response to treatment would be more evident and long-lasting in Grade 1 AKs and that even though some features were resolved at 3 months, they could reappear by 6 months. Similarly, Campione et al.30 stated that reduction of hyperkeratosis and parakeratosis were the first sign of response, followed by the normalization of the honeycomb pattern. On the other hand, Bandeira de Melo Cavalcante et al.41 found a poor agreement between RCM and histology after treating AKs with PDT (56.3% for atypical cells; 25% for dysplasia; no agreement for elastosis). The authors hypothesized that this could happen due to two different people evaluating the results of the techniques.

Besides, Pasquali et al.46 studied the efficacy profile of the combination of cryosurgery and ingenol mebutate in AKs, showing no difference according to the order of administration of both treatments. Nonetheless, a milder inflammatory skin reaction was seen when ingenol mebutate was administered first, suggesting this regimen as the preferred option due to a better tolerability.

Regarding treatment of BCC with ablative laser, it should be stated that most of the studies lacked a follow-up, as they only assessed effectiveness of the therapy immediately after performing laser. Sierra et al.52 suggested that the two recurrences observed in their study may have resulted from residual BCCs involving deep structures that were not detectable by RCM, leading to suboptimal outcomes with RCM-guided laser ablation.

Regarding RCM-monitorization of BCCs treated with cryosurgery, in the study of Ahlgrimm-Siess et al.,47 RCM underestimated the effectiveness of cryosurgery vs histology, suggesting that the scarring process may interfere with RCM image interpretation. Moreover, these authors highlighted that histologic clearance only occurred in those 8/10 lesions showing an early dermal reaction at RCM (bright round-polygonal dermal structures at 5h vs the additional 2/10 developing these changes at 24h), suggesting a possible response predictor starting at early evaluations. On the other hand, Pasquali et al.55 suggested that the slowly-progressive disappearance of BCC features may reflect the time needed by the immune system to activate and remove tumor cells. They also raised awareness of the possibility of pseudocysts being mistaken for tumor nest persistence.

Regarding IMQ, a case series by Zou et al.9 reported the presence of dendritic structures and bright epidermal particles at week 12. Similarly, Sahu et al.57 reported that an increased inflammation with scarce vasculature in RCM could predict a better response to IMQ. These structures may be predictors of response.

Regarding PDT, of note, two studies on RCM showed subclinical persistence of BCCs, allowing further treatment and thus avoiding recurrences.1,7 Similarly, Longo et al.1 raised awareness of the different appearance of residual BCCs after treatment with PDT compared to non-treated BCCs, as blurred or faint tumor islands could represent the only findings tumor persistence vs the dark silhouettes seen in non-treated BCC. These authors also reported that the activated LC and inflammatory cells would already appear one week after PDT lasting 1 month, representing a transient activation of the immune system.1

Finally, Banzhaf et al.59 studied the efficacy profile of combining ablative laser with ingenol mebutate, reaching higher complete response rates for superficial BCCs vs nodular BCCs. In this work, RCM was able to detect subclinical BCC persistence. Even though it did not reach statistical significance, the authors found a tendency of a better response to 0.05% of IM vs 0.015%.

In squamous tumors, RCM evaluation within the first 4 weeks after treatment showed an inflammatory response with the presence of dendritic cells supporting the hypothesis that the immune system may take several weeks to eliminate tumor cells.2,4,13,23,27,31 Similarly, in BCCs, an RCM evaluation within the first 6 weeks after treatment also showed inflammation, including abundant Langerhans cells, and persistence of tumor nests with acute changes such as necrotic basal cells or nest detachment.1,5,47,51,53,55,58,61 Tumor persistence may still be observed by 3 months,9,34,47,58,59,61 while at a later point in the follow up (e.g., 12–18 months) no signs of tumor can be seen in patients with complete response.1,7,9,61 Thus, even though more evidence is needed, it is suggested that RCM complete response assessment is conducted starting at least 4 weeks into therapy in squamous conditions, especially AKs, and starting at least 3 months into therapy in BCCs.

Regarding predictor markers of response, in case of AKs, it has been stated that the first sign is the reduction of hyper and parakeratosis, followed by the normalization of the honeycomb pattern.15,30,33 In BCC treatment, an early evaluation after cryosurgery, may help predict response, as the presence of a dermal reaction within the first 5h may predict a better response, compared with apparition of this reaction by 12h.47 Features suggesting a better response to IMQ can also be seen at later points during follow up, such as increased inflammation and bright epidermal particles.9,57 During follow-up, recurrence of treated BCCs may appear as blurred islands, compared to the dark silhouettes seen in non-treated BCCs.1 In addition, it should be highlighted the importance of long follow-up periods as there might be late recurrences.52,58

First, most of the identified studies presented risk of bias in more than one of the assessed items. Second, follow-up periods and treatment regimens were heterogeneous between studies. Nonetheless, combining the available results some suggestions could be made regarding follow-up time of non-invasively treated lesions with RMC. Thus, even though some suggestions were presented herein, larger studies with homogeneous protocols are needed to reach more robust recommendations.