Abstract
While most individuals with a clinical diagnosis of Neurofibromatosis type 1 (NF1) have a detectable pathogenic variant in the NF1 gene, other conditions have phenotypic features overlapping with NF1. Without molecular confirmation, individuals may be misdiagnosed and have a different underlying condition. Namely, if a child has constitutional mismatch repair deficiency (CMMRD), early detection and prevention strategies for cancer risk would include surveillance recommendations not typically recommended for children with NF1. This study aimed to explore phenotypes of individuals with a clinical diagnosis of NF1 to identify subpopulations who may benefit from further genetic counseling or testing for an alternate diagnosis. Retrospective review of 240 medical records of children who attended a neurocutaneous clinic identified 135 children with a molecularly confirmed pathogenic variant in NF1 or autosomal dominant pattern of clinical NF1 (“controls”) and 102 children deemed “at-risk” for another condition like CMMRD. Clinical presentation, family history of NF1, personal history of cancer, and family history of cancer were compared. When comparing clinical presentation, family history, and cancer history, minimal statistical differences were found, indicating that the at-risk population appears clinically indistinguishable from those with a clear diagnosis of NF1. Given the lack of distinguishable features between the at-risk and control population, this study suggests that tiered genetic testing for all individuals being evaluated for NF1 may be beneficial for identifying patients who may be misdiagnosed with NF1 and subsequently mismanaged. This study suggests that at-risk population with a suspected NF1 diagnosis may benefit from further evaluation. Correct diagnosis of constitutional mismatch repair deficiency is crucial to diagnose cancer at an early stage or prevent cancer from occurring.
Prevention relevance: This study suggests that at-risk population with a suspected NF1 diagnosis may benefit from further evaluation. Correct diagnosis of constitutional mismatch repair deficiency is crucial to diagnose cancer at an early stage or prevent cancer from occurring.
Introduction
Neurofibromatosis type 1 (NF1) is an autosomal-dominant genetic condition with variable expressivity, penetrance, and an incidence of approximately 1 in 2,500–3,000 individuals worldwide (1). The de novo rate is approximately 50% (2). In 1987, the National Institute of Health (NIH) Consensus Conference agreed on a defined list of clinical features that allow a medical professional to clinically diagnose NF1 (1). A clinical diagnosis of NF1 requires patients to meet at least two defined criteria from a list of seven features (2). These features include:
6 or more café au lait macules
Axillary or inguinal freckling
2 or more Lisch modules
2 or more neurofibromas of any type or 1 or more plexiform neurofibromas
A distinctive osseous lesion such as sphenoid dysplasia of tibial pseudoarthrosis
1 first degree relative with a confirmed diagnosis of NF1.
Alternatively, some individuals who are suspected to have NF1 may undergo genetic testing to determine whether they carry a pathogenic variant in the NF1 gene, regardless of whether clinical criteria are met. Identification of a pathogenic variant in the NF1 gene gives an individual a molecular diagnosis of NF1.
Historically, individuals that had been clinically diagnosed with NF1 had not been offered or opted not to pursue molecular testing for NF1 as it would not change their medical management. It is estimated that molecular testing can identify a pathogenic variant in the NF1 gene in patients that meet clinical criteria for NF1 approximately 95% of the time. It has been assumed that the remaining 5% of individuals do not have a pathogenic variant detected in NF1 due to the sensitivity of the test and its inability to detect all variants in NF1 (2). A portion of the 5% with a negative test result is also attributed to the possibility of mosaicism in these individuals. This may occur up to 10% of the time in an individual who is the first person in the family to have features of NF1 (3). While sensitivity is improved with RNA testing, some pathogenic variants are still missed, and so individuals who meet NF1 clinical criteria are managed as such regardless of molecular NF1 status (4).
Although NF1 is one of the more common genetic conditions associated with café au lait macules and neurofibromas, there are several other genetic conditions that have similar features. These conditions include, but are not limited to: Legius syndrome, McCune–Albright syndrome, LEOPARD syndrome, and Constitutional Mismatch Repair Deficiency (CMMRD; ref. 5). Although several of these syndromes have clear clinical features distinguishable from NF1, others do not. Furthermore, several reports have identified individuals who had been clinically diagnosed with NF1 prior to developing a cancer that is uncharacteristic of NF1. Notably, some of these individuals, upon further investigation, were reassigned the diagnosis of CMMRD (6, 7). Although CMMRD is rare even in individuals that may have features of NF1 (about 0.41% of individuals who are suspected to have NF1 may actually have CMMRD according to a recent study completed by Perez-Valencia and colleagues; ref. 8), the impact of an incorrect diagnosis could be devastating to these individuals and families. The emerging concern in cancer genetics is that individuals with a clinical diagnosis of NF1 who truly have CMMRD may miss an opportunity to prevent or detect a cancer diagnosis earlier with cancer surveillance if not identified properly (9, 10).
The purpose of this study is to help identify patients who may benefit from further evaluation of alternate diagnoses by characterizing a clinically diagnosed NF1 population. This study evaluates individuals with a clinical diagnosis of NF1 or those who are being followed as if they have a diagnosis of NF1, to determine whether there is an at-risk population that may be misdiagnosed. Individuals would therefore be managed incorrectly because of this incorrect diagnosis. Although the recent study by Perez and colleagues examines a similar population, our study attempts to further delineate those who may benefit from additional genetic testing by exploring subpopulations based on clinical findings. Identifying at-risk individuals in an effort to provide the correct diagnosis could be lifesaving, especially for those who are actually affected by CMMRD (9).
Materials and Methods
Chart review
A retrospective chart review was completed for patients who had been evaluated at Children's Hospital Colorado (CHCO) for NF1. All individuals had been given a diagnosis of NF1 or are being followed due to suspicion of NF1. An I2B2 search was completed to identify charts for review. I2B2 is a CHCO electronic medical record integrated search engine used to capture large populations of individuals meeting certain inclusion and exclusion criteria as determined by the investigator. For this study, the inclusion and exclusion criteria were as follows:
Inclusion criteria:
(i) ICD10 codes of Q85.00 (neurofibromatosis, unspecified), Q85.01 (neurofibromatosis, type 1), Q85.09 (other neurofibromatosis) or Q85.0 (neurofibromatosis (nonmalignant), or the ICD9 codes of 237.70 (neurofibromatosis, unspecified), 237.71 (neurofibromatosis, type 1), or 237.79 (other neurofibromatosis)
(ii) The patient must have been seen by the Neurocutaneous Genetics clinic or the NF1 Multidisciplinary clinic
(iii) The patient must have been between the ages of 31 days to 18 years at time of I2B2 search.
Exclusion criteria:
(i) ICD-10 code Q85.2 (neurofibromatosis, type 2) or the ICD-9 code 237.72 (neurofibromatosis, type 2)
(ii) Individuals initially evaluated in one of the clinics mentioned above after August 12, 2016 which is the date the study protocol was approved by the Colorado Multiple Institutional Review Board (COMIRB) and charts must be retrospectively reviewed from this date.
(iii) Patients with other genetic disorders, namely CMMRD, are likely to develop cancer before age 18. Therefore, patients over age 18 were excluded from this study.
Charts were reviewed by date of patient's last visit from the most recent date backwards, until 240 cases that met criteria were reached. This number was selected for timely completion of the project within a Master's program. Patient visits between September 9, 2014 and August 12, 2016 were reviewed.
Upon initial review of the patient charts, patients were excluded if no pedigree was available or if there was no documentation by a geneticist and/or a genetic counselor on the date of the encounter. In addition, individuals who had a pathogenic variant in SPRED1 or a mismatch repair gene would have been excluded, however, no individuals reviewed had a pathogenic variant reported in any of these genes. This study was conducted in accordance with U.S. Common Rule guidelines for the ethical principles for research involving human subjects, though the Colorado Multiple Institutional Review Board (COMIRB) review deemed the research to have minimal risk. The study was a retrospective chart review and did not require patient contact, and so, written consent was not obtained and COMIRB approval was expedited.
Data collection
Individual patient medical records were reviewed, and relevant data were recorded in a REDCap database. Data collected included variables regarding demographics, clinical features of NF1, personal history of cancer, family history of cancer, and supplemental information (11–21). A full list of the data collected is outlined in Appendix A and B. All identifying information collection was de-identified when retrieving the raw data.
Data analysis
Raw data were extracted from REDCap into Microsoft Excel after 240 cases were reviewed. Patients were separated into three different groups: (i) controls: individuals who received genetic testing and tested positive for an NF1 pathogenic variant; (ii) individuals who received NF1 genetic testing and did not have a pathogenic variant in NF1 detected; and (iii) individuals with a clinical diagnosis of NF1, or suspected diagnosis of NF1, who have not pursued molecular testing for NF1. Groups 2 and 3 were combined and considered the at-risk population with the exception of individuals whose family history was consistent with an autosomal-dominant pattern of inheritance of NF1. The individuals with family histories consistent with autosomal dominant NF1 were included in the “control” population regardless of molecular status. For the purposes of this study, an individual was only counted as having an autosomal-dominant family history if the person's mother or father reported a clinical diagnosis of NF1 or a known familial mutation was disclosed (Fig. 1).
Flow chart of how the control and at-risk populations were determined. The middle grey boxes are the at-risk population, and the dark gray boxes are the control population.
The control and the at-risk populations were compared to one another by features of NF1, personal history of cancer, and family history of cancer. Features of NF1 were separated into six clinical categories. Individuals were considered to have NF1 features from each clinical category if they had the minimum number of clinical features in that category. For example, an individual was assigned a score of 0 if they had 0–1 dermatologic features and were assigned a score of 1 if they had 2 or more dermatologic features (see appendix A for list of clinical features and explanation of scoring for each clinical category). A total clinical features score between 0 and 6 was assigned to each patient by adding up each clinical category for which they met criteria. Cancer types were grouped into 9 categories to compare the control and at-risk populations: skin, blood, gastrointestinal, urinary tract, female reproductive tract, male reproductive, central nervous system (CNS), and breast cancers.
Presence or absence of features in each category were compared across the control group and the at-risk group using a t-test (apart from cancer history) and a Fisher exact test to determine whether there were differences between groups. Family history of cancer included histories from first- and second-degree relatives, but more distant relatives' histories were excluded because of inconsistency of reporting.
Data Availability
Raw data included as Supplementary File.
Results
The I2B2 search identified a total of 682 individuals that met criteria. A total of 240 charts were reviewed using the study methods. Three patients who had a variant of uncertain significance detected in NF1 were excluded from the final study population because they did not fall into the control or at-risk population.
Demographics
The majority of patients identified as white/Caucasian, although almost 40% of individuals reported being Hispanic/Latino. Average age of patients was 8.3 years (ranging from 6 months – 17 years). 92.59% (n = 125) of individuals in the control population met clinical criteria for NF1, while 86.27% (n = 88) of individuals in the at-risk population met clinical criteria for NF1. Table 1 contains a summary of the study population demographics.
Distribution of ethnicity in the total population studied.
Genetic testing for NF1
Of the 240 individuals analyzed, 133 had not undergone genetic testing for NF1. Of those tested, 82 had a pathogenic variant detected in NF1, 22 had no variants detected in NF1, and 3 had a variant of uncertain significance detected in NF1 (Fig. 1). SPRED1 genetic testing was also pursued in 49 of the 82 individuals who tested positive for a pathogenic variant in the NF1 gene. While 19 of the 22 individuals who tested negative also had SPRED1 testing. No individual in the study population had a pathogenic variant detected in the SPRED1 gene. After identifying individuals with our defined autosomal-dominant family, the control group contained 135 individuals and the at-risk group contained 102 individuals. Although the genetic testing methodologies were not captured in this dataset, the majority of individuals who pursue genetic testing for NF1 and SPRED1 though CHCO receive testing through the University of Alabama (UAB), an academic laboratory experienced in testing for these syndromes.
Features of NF1
t test comparisons between the control and at-risk groups did not detect statistically significant differences when comparing total clinical features score. Interestingly, when evaluating for general trends based on age of the individual and their clinical features score, the control group followed the general trend expected, namely more clinical features suggestive of NF1 as age increased until approximately 8–10 years of age. The at-risk group, however, did not appear to follow this trend nor any particular pattern with age.
Each clinical category was analyzed independently to compare the control and at-risk populations. There were no statistical differences between the at-risk and control populations when quantifying the number of clinical features in each category, with the one exception being the neurologic features category. The control population demonstrated a higher average number of neurological features than the at-risk population (P = 0.0073; Fig. 2). Interestingly, the incidence of an individual having one or more osseous lesion specifically was about 3–4 times higher in the at-risk population than in the control population (4% and 1%, respectively). None of the individuals with an osseous lesion in the at-risk population had undergone genetic testing.
Clinical features score comparison. There is a statistically significant difference between the control population and the at-risk population only in the neurologic category (P = 0.0073).
Finally, when the at-risk population was further delineated into two subpopulations: (i) individuals who did not have genetic testing and (ii) individuals who tested negative for NF1 pathogenic variants, significant differences were observed in several clinical categories. Statistically significant differences were found for dermatological features, ocular features, and endocrine and growth features (P = <0.0001, P = 0.0114, P = 0.0024, respectively). For all three clinical feature groups, the negative genetic testing population had significantly fewer clinical features (Fig. 3).
Comparison of clinical feature subpopulations within at-risk group. Statistically significant differences seen in dermatologic, ocular, and endocrine and growth categories (P ≥ 0.001, 0.0114, and 0.0024, respectively).
Family history of NF1
Of the 22 individuals that had negative genetic testing, 3 individuals reported a family history of NF1. Upon further evaluation, only 1 had a positive family history that demonstrated a clear autosomal-dominant pattern. The remaining 2 individuals had an affected sibling only or a diagnosis of NF1 that appeared to skip generations. Thus, a confirmed pattern of autosomal-dominant inheritance could not be confirmed in 21 of the 22 patients who underwent genetic testing but had no pathogenic variants detected. Given that all patients with an autosomal-dominant pattern of NF1 were included in the control group, and all without were in the at-risk group, no comparisons between the groups regarding family history of NF1 were made.
Personal history of cancer
Eleven individuals in the study had a personal history of cancer. Of those, 8 individuals had a pathogenic variant detected in NF1, and 3 had not undergone molecular testing for NF1. Individuals in the control group had histories of astrocytoma, unspecified brain tumor, low-grade glioma, rhabdomyosarcoma, lesion near the pituitary, and “thalamic mass”. Individuals who had not undergone genetic testing had histories of low-grade glioma (n = 3). No personal histories of cancer were reported in any of the patients who had genetic testing with no variants detected. No significant differences between the control and at-risk populations were observed when grouping together all cancer types (P = 0.3592). Although not statistically significant, the control population had 3.38% of individuals with a reported history of cancers or tumors while the at-risk population had 1.27%.
Family history of cancer
The presence of cancer in at least 1 first- or second-degree relative was not statistically significant between the control and at-risk populations. In addition, there were no statistical differences detected when comparing individuals with a positive family history of at least 2 first- or second-degree relatives. Interestingly, more individuals in the at-risk population (33.33%, n = 12) had more than one affected family member with cancer than in the control population (20.75% n = 11). The types of cancers in both the control and at-risk populations were similar, with the exception of more CNS tumors in the control group (22.86%; n = 16) compared with the at-risk group (7.41%; n = 3; P = 0.0261; Fig. 4).
Comparison of the control and at-risk populations when reviewing family members affected by cancer. Only statistical differences were in the CNS cancers category (P = 0.0261).
Of note, 7 types of cancer reported in the family members from the at-risk population that were not reported in the control population: acute lymphoblastic leukemia, endometrial cancer, neuroblastoma, esophageal cancer, lymphoma, osteosarcoma, and thyroid cancer.
Finally, when comparing the two subpopulations within the at-risk population (negative genetic testing vs. no genetic testing), more (P = 0.0484) of the patients who did not have genetic testing (55.56%, n = 45) had a family history of cancer than those who had genetic testing with no pathogenic variant detected (28.57%, n = 6).
Discussion
This study serves as a preliminary descriptive analysis of the NF1 population at one large academic pediatric hospital with respect to identifying individuals with clinical features of NF1 who may actually be affected a different underlying genetic etiology. Clinical differences between the control and at-risk populations were limited, suggesting that relying on clinical phenotype alone to differentiate individuals with NF1 versus another syndrome is insufficient. Regarding family history of cancer, though the at-risk population appeared to report a family history of cancer more often than the control population, there was not a statistically significant difference between the two populations. Individuals who may have an underlying diagnosis of CMMRD most often have biallelic pathogenic variants in either PMS2 or MSH6 mismatch repair genes (10). Classically, individuals with a monoallelic mutation in the PMS2 or MSH6 genes may not have a striking personal or family history of cancer given the low penetrance. Given these facts and findings, family history of cancer may not be a reliable way to distinguish the at-risk population for a condition like CMMRD.
Given the lack of differences between the control and at-risk populations across all clinical categories analyzed in this study, it appears the most effective way to identify individuals that may have a different underlying etiology for their presenting features may be to pursue molecular analysis via a panel including genes associated with multiple genetic syndromes or offering genetic testing in a step-wise fashion as indicated by Suerink and colleagues (3). As outlined in the suggested criteria for diagnosing a child with CMMRD before a cancer diagnosis, Suerink and colleagues states completing comprehensive genetic testing for NF1 and SPRED1 before moving on to alternate diagnoses such as CMMRD emphasizes the importance of molecular confirmation of a clinical diagnosis or suspicion of NF1 or Legius syndrome (3). Historically, confirming a clinical diagnosis of NF1 by pursing genetic testing was driven by personal choice. However, evidence strongly supports that genetic testing is important to confirm a clinical diagnosis so that the patient can be monitored appropriately for the correct underlying genetic condition.
Limitations
This study has several limitations including the possibility of limited generalizability as all individuals in this study were seen in one institution. Other centers may have different referral practices or different practices surrounding genetic testing for this population. These practices also change with time and data from these patients has not been updated since 2016; however, there should still be a subpopulation that may benefit from further evaluation as not all individuals have pursued comprehensive genetic testing. This study was limited by the amount of information provided to the genetic counselor and physician at the time of visit. The detail of information given at the time of visit is variable from patient to patient, and there is no way of knowing if the information given was comprehensive. Age of onset was underreported for both features of NF1 and family history of cancer. In addition, this dataset did not distinguish between types of gliomas, which may provide more insight as to whether a glioma (if it were an optic glioma) would be more indicative of a NF1 diagnosis.
Although 240 charts were reviewed, only 22 individuals were found to be negative for pathogenic variants in NF1 gene. We were unable to further delineate the population that did not undergo genetic testing after considering an individual's family history. Over half of this population was included in the at-risk population. Some of these individuals may be affected by NF1 and were the first individual in their family to have it, but this could not be distinguished. This could have skewed the data for the at-risk population. In addition, individuals with negative genetic testing may have mosaic NF1 which is undetectable in a fibroblast sample, resulting in a negative test result. There may also have been human error while documenting the individuals' information into their electronic medical record or while extracting that information and entering it into the database for analysis. In addition, the I2B2 software may not have detected all eligible patients for the study. Data may also be skewed due to sibships included in this study. There were 8 known sibships and 4 known half sibships.
Finally, this study did not assess barriers to genetic testing in the at-risk population. For example, individuals in the at-risk population may have been interested in pursuing genetic testing but did not pursue testing given the cost. Historically, genetic testing for NF1 had a high cost. Although one of the testing laboratories that offers specialized comprehensive genetic testing for NF1 is also still the most expensive, there are now many alternate genetic testing laboratories that offer sufficient coverage of the NF1 gene at a greatly reduced price. These labs also include analysis of genes associated with possible alternate diagnoses, and at lower costs.
Future research
An area of future study could be to focus on the population that has not had genetic testing for NF1 but are being followed medically as if they are affected by NF1 based on clinical features. This could further narrow who would benefit from discussion and evaluation for an alternate diagnosis if they move forward with genetic testing. This may help further define the at-risk population as we were limited to include most individuals that did not undergo genetic testing in the at-risk population with no clear way to determine who did or did not have NF1.
In addition, further analysis on the type of genetic testing that was received historically and through which testing lab would be an important area of future study. If individuals received genetic testing through a testing lab that does not offer comprehensive coverage of the NF1 gene, individuals may consider being retested at an alternate testing lab. This would further define the at-risk population as well.
Exploring age of onset of the development of clinical features could also be informative and further define when the development of these features may plateau. If an individual is not following the general trend seen in populations confirmed to have a molecular diagnosis of NF1, an alternate diagnosis could be considered.
Another area of future research for the at-risk population identified by this study is to provide additional testing using a process such as the proposed criteria by Suerink and colleagues This would help further delineate the subset of individuals in the at-risk population most likely to be at risk of having a diagnosis of another condition like CMMRD. The individuals in the at-risk population could be offered genetic testing either by a comprehensive next generation sequencing panel which would include genes that cause many of the known underlying genetic conditions with overlapping features, or they could be offered testing in a step-wise approach (22). If testing was offered as a comprehensive panel, special ethical and psychosocial considerations should be made for the possibility of identifying Lynch syndrome, rather than CMMRD given the commonality of identifying one pathogenic variant in a mismatch repair gene rather than two. Depending on the clinical features, mismatch repair protein IHC staining may also be a consideration for individuals in the at-risk population.
Alternatively, this at-risk population could first undergo comprehensive NF1 and SPRED1 testing and individuals who test negative could have MSI studies completed could help identify individuals that may benefit from genetic testing for CMMRD. This would help avoid incidentally diagnosing children with Lynch syndrome, an adult-onset condition and which is significantly more common than CMMRD. This was proposed by a recent study completed by Perez-Valencia and colleagues to identify individuals who had been misdiagnosed with NF1 without pursing excessive genetic testing to avoid incidental diagnoses.
Conclusion
This study provides an initial look at the complexity of properly diagnosing individuals with NF1 versus other diagnoses with overlapping clinical features of NF1. This emphasizes the importance for families to pursue genetic testing to confirm a diagnosis of NF1 rather than relying on a suspected diagnosis based on presenting clinical features. Especially for children, an alternate genetic diagnosis could significantly change the clinical management of that child. In the case of CMMRD, a correct diagnosis is crucial to prevent a cancer diagnosis or detecting cancer development at an early stage to provide the best prognosis possible. Genetic counselors and other medical professionals caring for these families may consider explaining that though NF1 may be a common genetic syndrome associated with the clinical features seen in the individual, genetic testing could be utilized to confirm this diagnosis. These providers could further review that genetic testing may also indicate that an alternate diagnosis is the cause of the NF1-like features, which would drastically alter medical management.
Authors' Disclosures
K.W. Schneider is a member of the International Replication Repair Deficiency Consortium. No disclosures were reported by the other authors.
Authors' Contributions
S.R. Hicks: Conceptualization, data curation, formal analysis, investigation, methodology, writing–original draft, writing–review and editing, this author completed this research as part of a requirement for her Master's program. The majority of this study was completed by Ms. Hicks. A.K. Cozart: Conceptualization, resources, supervision, writing–review and editing. G.A. Bellus: Conceptualization, resources, supervision, writing–review and editing. K.W. Schneider: Conceptualization, resources, data curation, supervision, validation, project administration, writing–review and editing.
Acknowledgments
This research study was completed as part of a Master's degree in genetic counseling training requirement. We would like to thank all of the healthcare providers and patients followed in the neurocutaneous clinic at Children's Hospital Colorado. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Footnotes
Note: Supplementary data for this article are available at Cancer Prevention Research Online (http://cancerprevres.aacrjournals.org/).
Cancer Prev Res 2021;14:471–8
- Received July 24, 2020.
- Revision received November 3, 2020.
- Accepted January 5, 2021.
- Published first January 11, 2021.
- ©2021 American Association for Cancer Research.
References
Volume 14, Issue 4
