Cutaneous neonatal Langerhans cell histiocytosis: a systematic review of case reports [version 1; peer review: 1 approved with reservations, 1 not approved]

Background: Cutaneous langerhans cell histiocytosis (LCH) is a rare disorder characterized by proliferation of cells with phenotypical characteristics of Langerhans cells. Although some cases spontaneously resolve, no consistent variables have been identified that predict which cases will manifest with systemic disease later in childhood. Methods: A systematic review (Pubmed, Embase, Cochrane database and all published abstracts from 1946-2018) was undertaken to collate all reported cases of cutaneous LCH in the international literature. This study was registered with PROSPERO (CRD42016051952). Descriptive statistics and correlation analyses were undertaken. Bias was analyzed according to GRADE criteria. Results: A total of 83 articles encompassing 128 cases of cutaneous LCH were identified. Multiple lesions were weakly associated with an increased length of survival (R=0.304 (p<0.05)), Worse prognosis was associated with internal organ involvement with a statistically significant chi squared statistic (χ 2 =14.96, 2DF p<0.001) and an elevated odds ratio ((OR)= 12.30 95% CI=2.67-56.74). Vesicular lesions (OR=10.8 95% CI=2.83-41.26), but not ulceration (OR=0.53 95% CI 0.12-2.05) were associated with greater risk of mortality. Conclusions: Congenital and neonatal LCH most commonly presents as multiple lesions in multiple anatomical sites at birth. Significant differences, including the associations of mortality with lesion morphology and number were seen in this neonatal cohort compared to overall pediatric LCH. These findings require validation in a large prospective cohort. Dr. Venning and colleagues perform a review on neonatal and congenital cases of Langerhans cell histiocytosis (LCH) to examine predictors for progression to systemic disease. While this endeavour will help to risk-stratify patients, there are several issues that need to be addressed: ABSTRACT In the abstract, the text should make reference to congenital and neonatal onset LCH, as this is the focus of their research.


Background
Cutaneous Langerhans cell histiocytosis (LCH) is a rare disorder manifest in the proliferation of cells with phenotypical characteristics of Langerhans cells which involves the cutaneous structures 1 . We have used the term 'cutaneous' in this review to differentiate from 'skin-limited' which implies the absence of systemic disease involvement 1,2 . The incidence of cutaneous LCH varies from two to nine cases per million children per year 1,2 . Rarely, the disease is present at birth or in the neonatal period. A proportion of these cases spontaneously resolve however no consistent variables have been identified which provide predictive value as to which cases will resolve or remain skin-limited, and which will manifest with multisystem LCH later in life 1,2 . The rate of progression of cutaneous LCH to other organs has varied widely in previous studies, from 0 to 60% 1 . This lack of accurate and reliable data makes it difficult to provide information to patients regarding the risk of progression of disease and limits the development of evidence-based screening measures to identify the presence of systemic disease. Currently, consensus guidelines 1 state that most cases of cutaneous LCH spontaneously regress but some cases do progress to multisystem disease 1 . It is unclear whether cutaneous LCH is merely clinically more easily identified and hence often precedes diagnosis of internal disease. This would also suggest that widespread screening of cases of cutaneous LCH may produce leadtime bias in the survival rates of individuals with multisystem LCH with cutaneous involvement, an issue which to date has not been explored. Currently, in cases of cutaneous LCH screening is considered mandatory 1,2 .
Regarding identified risk factors for disease progression and mortality, overwhelmingly the data is sourced from cases of systemic LCH 3,4 , which may or may not include cutaneous disease. Data from older children also far exceeds data from neonatal cohorts, limiting or knowledge of differences between presentations in the neonatal population and older pediatric age groups. Only one retrospective case series of 19 patients 5 examined survival outcomes in infants diagnosed with cutaneous LCH within the first 4 weeks of life, with long-term follow-up beyond 10 years being limited to small case series of less than 10 patients 6 . In the setting of systemic LCH, inadequate response to initial therapy and risk organ involvement, (defined as bone marrow, liver, spleen and/or lung), are the currently associated with adverse clinical outcomes and mortality in LCH 1,2 .
Isolated bone involvement portends significantly prolonged survival compared with other organ involvement 3,4 . As expected, patients with multiple organ involvement have been found to have the highest risk of progression and mortality 7 . Detection of the BRAF-V600E mutation (often seen in systemic LCH but rarely in skin-limited LCH), has also been associated with increased risk of disease recurrence 4 .
Overall, given the reports (albeit uncommon) of progression of cutaneous LCH to multisystem disease, the identification of clinical or histological predictive variables may reduce rates of unnecessary invasive screening in neonates with skin-limited LCH.

Objectives
To collate all published cases of cutaneous congenital/neonatal LCH.
To perform a descriptive analysis of cases and reviews to evaluate mortality risk and risk of progression to systemic disease.
To identify risk factors which may contribute to mortality risk and risk of progression to systematic disease.

Methods
This systematic review was registered with PROSPERO (Registration number CRD42016041425) and was conducted in line with the PRISMA statement 8

Data sources
Information Sources for this review encompassed Medline (1946-March 1 2018), Embase (1980-March 1 2018) as well as "Epub ahead of print, and non-indexed citations" as shown in Figure 1. The search strategy is presented in Study eligibility criteria Eligibility criteria for this review included published case reports, case series and reviews with no restrictions of patient sex or ethnicity and language of publication. Eligible cases included: 1) Cases of histologically diagnosed LCH at birth (congenital) or within the first 4 weeks of life involving the skin.
2) Cases which report data pertaining to evidence of systemic involvement (clinical examination, skeletal survey etc.) and/or histological data (CD1a, eosinophil density etc.) 3) Cases with follow up data of any period.

Appraisal and synthesis methods
Data collection was performed independently by two independent authors (EH and EY), with any disagreements regarding inclusion of citations being referred to a third author (VV) for mediation. Information was collected using a standardized data collection form (available as extended data on OSF 9 ) with the principal outcomes of interest being mortality, age at demise and length of follow up. Data not available from the published article was requested via email contact with the relevant corresponding authors. Potential sources of bias in collating cases were acknowledged including publications bias and reporting bias regarding the overall incidence of congenital and neonatal LCH, therefore only cases with a diagnosis of cutaneous LCH at birth (congenital) or within the first 4 weeks of life (neonatal) were included, and no attempt to quantify the number of cases of systemic LCH with a "missed" diagnosis of self-resolving cutaneous congenital LCH was undertaken. Particular effort was made to include unpublished cases and cases presented as posters and abstracts in order to reduce the impact of publication bias in our analyses.
An exploratory univariate analysis (using Pearson correlation coefficients for categorical variables and chi-squared tests for binary variables) was undertaken to correlate mortality and the progression to systemic disease with the clinical and histological variables collated.

Study selection
A total of 211 articles were identified in the literature review; 82 of these articles were removed upon review of titles and abstracts against eligibility criteria. Full-text review of 129 articles excluded 12 review articles, 1 duplicated case report and 33 articles (containing 42 cases) due to lack of follow up data. The remaining 83 articles 5,10-91 containing 128 individual cases were used as the basis of this review.

Summary of findings
The summarized demographic data of the included cases is presented in Table 2.

Univariate correlation analysis
The results of univariate correlation analysis are summarized in Table 3 was 27.4% (n= 17/62). The calculated OR for mortality based upon the presence of systemic involvement was 12.3 (95% CI). No statistically significant associations or OR were seen between histological markers and clinical outcomes including mortality or length of survival in the data examined. Given the heterogeneity of the sample, no multivariate analysis was performed on the collated data.

Summary of evidence
The results of this systematic review of case reports of cutaneous neonatal LCH differ from the pre-existing literature in several areas. This may be because existing data includes all cases of pediatric LCH, as opposed to the congenital and neonatal cases focused on in this review. This highlights the need for recognition that congenital and neonatal LCH have inherently different clinical characteristics compared to other pediatric cases of LCH. Minkov et al. 3 have reported that the trunk was the most common overall site of disease. However, our data suggest that a large proportion of congenital and neonatal cases involve multiple anatomical sites (n=65). No significant gender predominance was identified (males=63; females=55). A weak association was seen between a later onset of disease and a worse prognosis (r=0.263, p<0.05). This is in line with the literature with earlier onset disease significantly associated with spontaneous resolution 92,93 .
Systemic disease was identified in 48.4% of cases (n=62) lower than the rates for the overall pediatric group at 59% 3 , and those Regarding lesion morphology, Battistella et al. 92 suggest that single, necrotic, hypopigmented macules and distal topography (lesions present at a distal site) suggest a self-regressive form of disease 92 . This is still an area of contention with no reliable data from cohorts larger than 20 patients 5,93,95-98 . We identified a weak correlation between skin lesion descriptors and overall mortality as well as length of survival in the neonatal and congenital LCH population. The presence of multiple lesions was associated with increased length of survival, although the presence of lead-time bias was likely given the non-significant differences in mortality between the two groups. Vesicular lesions were associated with increased mortality whereas no impact of survival was seen in the presence of ulcerated lesions. We anticipated that reporting bias would result in confirmation of an association between ulcerated lesions and mortality if one existed, although this has not been confirmed by our review. One explanation is that vesicular lesions commonly progress to ulceration during the stages of healing, thus emphasizing the need for consistent descriptors in case reports of LCH. Alternatively, ulcerated lesions might have an association with mortality but not in the congenital and neonatal LCH cohort.

Limitations
Most cases identified were congenital (67.9%; n=87), although some controversy exists regarding whether congenital cases exist at all 93 . Morren states that LCH presents prior to 3 months of age but does not occur congenitally 93 . Given the retrospective nature of our study, we were unable to shed further light on this debate as we were reliant upon multiple authors' observations and recordings.
Given the variability in patient follow-up in this review, the current estimates of mortality risk are only valid until 18 months of age (the mean length of follow-up). The lack of long-term follow-up is the major reason why data is lacking regarding long term recurrence rates in neonatal LCH and thus our review is limited to conclusions regarding short-and medium-term outcomes.
Future research should expand upon this by analyzing longerterm outcomes. Haupt 1 has recommended a long-term followup of 5 years for patients, mirroring that of childhood cancer survivors. This is applicable even to both skin-limited LCH and systemic disease. Progression of skin-limited LCH to multisystem involvement is documented in the literature 1,5 .
We had a limited ability to identify statistically significant variables that contribute to LCH mortality due to limited follow-up in documented cases. The GRADE approach 99 to assessing the quality of evidence and strength of recommendations (available as extended data on OSF 9 ) shows the absence of control groups, and incomplete follow up. Long-term, prospective, multicenter collaborative studies needed to confirm the findings of this review and are important steps in characterizing the progression of neonatal LCH.

Conclusions
We present a systematic review of case reports of cutaneous congenital and neonatal LCH. The descriptive characteristics in this review significantly differ from descriptions of overall pediatric LCH, highlighting the clinical differences between these entities. Congenital and neonatal LCH most commonly presents in multiple anatomical sites at or shortly after birth, with the presence or absence of systemic involvement significantly impacting mortality. Further prospective, long-term multicenter collaborative studies are required to corroborate the results of this review.
Data are available under the terms of the Creative Commons Zero "No rights reserved" data waiver (CC0 1.0 Public domain dedication).

Grant information
The author(s) declared that no grants were involved in supporting this work. demonstrated that 4/12 (33%) of patients with isolated skin LCH progressed to multi-system LCH.

METHODS
It is unclear if the the study eligibility criteria requires all 3 criteria to qualify. Please elaborate. As well, please document reasons for exclusion. I note that the study by Lau et al. 2 has data on patients but was not included.

1.
It is unclear what the statement "no attempt to quantify the number of cases of systemic LCH with a "missed" diagnosis of self-resolving cutaneous congenital LCH was undertaken" means. Does this mean these cases were included or excluded? 2.

RESULTS
Please define "multiple". Is this more than 1 lesion or 10 and why was this number chosen? As before, the distinction between skin limited LCH and cutaneous LCH should be clear.

1.
Among the paper's objectives is: "to identify risk factors which may contribute to mortality risk and risk of progression to systemic disease.", however, when I finished reading the paper, I still did not have a clear idea of what risk factors contribute to the mortality and progression risk in the congenital/neonatal subset of LCH. In the conclusion they state that the presentation of congenital/neonatal LCH more often involves multiple sites compared to paediatric LCH, and that systemic involvement significantly impacts mortality. To me, the latter point holds in both paediatric and congenital/neontal LCH and isn't new information, nor helpful in distinguishing the two.

LIMITATION
The statement "Morren states that LCH presents prior to 3 months of age but does not occur congenitally" is odd, as Morren's report lists 2 congenial cases 3 .

1.
Overall, this is an interesting undertaking. However, methodologically, there a number of issue that should be examined.
truly equivalent?
Certain components of the authors' interpretation need to be addressed. For instance, it is unclear why the authors preferred to study the influence of multiple lesions on length of survival, rather than on progression to systemic disease or mortality, especially considering the limitation of short follow-up period. Also, the authors conclude that the descriptive characteristics in this review significantly differ from descriptions of overall pediatric LCH. However, they do not provide a statistical comparison or clarification of the age range that they consider pediatric. It is also questionable whether meaningful analysis of the impact of lesions type can be performed considering that 31/128 (24%) of lesions were listed as "others" and another 9 were missing.
In the last paragraph of the results, the authors state that no statistically significant associations or OR were seen between histological markers and clinical outcomes. How could such a conclusion be drawn without central pathology review? In the view of the reviewers, such a correlation cannot be made based on a description of the pathology findings, regardless of how well and detailed they are described, but needs a review of the biopsy specimens by experienced pathologists in a structured way. Additionally, the authors report that they were unable to perform a multivariate analysis due to data heterogeneity. If by heterogeneous the authors were implying that the data was inconclusive and/or insufficient, then this poses questions to the findings of the univariate analysis either.
The following components should also be addressed: In the 1 st paragraph of the background, the authors write: "Cutaneous Langerhans cell histiocytosis (LCH) is a rare disorder manifest in the proliferation of cells with phenotypical characteristics of Langerhans cells which involves the cutaneous structures 1 ." However, it should be noted that Langerhans cells are dendritic cells of the skin and mucosa, and that despite their phenotypic resemblance to Langerhans cells (and shared immunohistochemical markers), the pathologic cells of LCH derive from immature myeloid precursor cells 1 . Additionally, reference one used by the authors is a set of guidelines. Accordingly, the original studies should be cited when statistics are noted (i.e. the incidence of cutaneous LCH and rate of progression). The authors write: "Currently, consensus guidelines 1 state that most cases of cutaneous LCH spontaneously regress but some cases do progress to multisystem disease 1 ". Do they mean "skin-limited" rather than "cutaneous" according to the distinction made by the authors that "skin-limited" implies the lack of systemic involvement? Though spontaneous remission is the norm for "skin-limited" LCH, skin-limited disease is rare (2% of cases) 2 . The authors write: "It is unclear whether cutaneous LCH is merely clinically more easily identified and hence often precedes diagnosis of internal disease". Actually, cutaneous disease is often misdiagnosed due to its resemblance to other more common conditions (i.e diaper dermatitis and cradle-cap) 3 . In addition to noting in the background section that screening for multisystem disease at the time of initial presentation is mandatory, the authors should also mention here the need for long-term follow-up due to the potential for disease reactivation following resolution or future progression to multisystem disease. The authors note this in the conclusion, but it should also be stated here. The authors only note one source in the conclusion giving this recommendation, which undermines its importance 4,5,6 .
In the 3 rd paragraph of the background, the authors write "Detection of the BRAF-V600E mutation (often seen in systemic LCH but rarely in skin-limited LCH), has also been associated with increased risk of disease recurrence 4 ". Of note, BRAF-V600E mutations are not only associated with recurrence, but also with treatment-refractory disease and permanent sequelae. These associations are observed in both isolated and disseminated LCH (and in disseminated disease are also associated with risk organ involvement) 7,8,9 . Furthermore, Héritier's study in a cohort of 315 patients with determined BRAF status conflict with the statement that BRAF-V600E is rare in skinlimited disease. They report the presence of BRAF-V600E mutations in 87.5% of patients with multifocal single system cutaneous disease and in 80.2% of patients with multifocal cutaneous multisystem disease. What might also be of interest to the authors is that the mutation was absent in the 6 infants with solitary cutaneous lesions and single system disease 10 .
In the 4 th paragraph of the background, the authors write: "Overall, given the reports (albeit uncommon) of progression of cutaneous LCH to multisystem disease, the identification of clinical or histological predictive variables may reduce rates of unnecessary invasive screening in neonates with skin-limited LCH". However, as we note above, multisystem disease in patients presenting with cutaneous involvement is the norm.
In the 1 st paragraph of the discussion, the authors write: "This is in line with the literature with earlier onset disease significantly associated with spontaneous resolution 92,93 ". In contrast, because high-risk multisystem disease has been negatively correlated with age, younger age is associated with a worse prognosis 11 . Subsequent to Minkov's findings, Gadner et al. reported no difference in treatment response between different age groups when correcting for the involvement of risk organ systems. Thus, the difference in prognosis between age groups is likely due to the higher prevalence of multisystem disease in younger patients 12 . Similarly, in the 2 nd paragraph, the authors write: "Systemic disease was identified in 48.4% of cases (n=62) lower than the rates for the overall pediatric group at 59% 3 , and those reported by Stein et al. (63.1%) 5 ". However, in a retrospective analysis of 61 neonates with LCH, Minkov et al. note a higher prevalence of multisystem disease in neonates (ironically, 59%) 11 .
The benefits of publishing with F1000Research: Your article is published within days, with no editorial bias • You can publish traditional articles, null/negative results, case reports, data notes and more • The peer review process is transparent and collaborative • Your article is indexed in PubMed after passing peer review • Dedicated customer support at every stage • For pre-submission enquiries, contact research@f1000.com