Comparison of Cancer Prevalence in Patients With Neurofibromatosis Type 1 at an Academic Cancer Center vs in the General Population From 1985 to 2020

This cohort study examines the cancer prevalence and survival outcomes among patients with neurofibromatosis type 1 vs the general population.


Introduction
Neurofibromatosis type 1 (NF1) is an inherited autosomal-dominant disorder that occurs in 1 in 3000 individuals. 1 Manifestations of NF1 can be associated with outcomes in every organ system. 2 Patients with NF1 have wide phenotypic variability, and the condition is associated with higher rates of benign and malignant tumors. [3][4][5] Life expectancy is 10 years to 15 years shorter than that among the general population, a decrease associated with malignant neoplasms. 6 Although an association with certain neoplasms in NF1 is well-recognized, the risk of specific neoplasms has been challenging to quantify accurately. 3 This study aimed to determine the prevalence, management, and outcomes among patients with NF1 who developed neoplasms other than neurofibromas and received treatment at a large multidisciplinary cancer center.

Methods
The University of Texas MD Anderson Cancer Center Institutional Review Board (IRB) approved this cohort study. The IRB granted a waiver of informed consent because the study is a retrospective medical record review with no more than minimal risk and the collected, limited data were double coded, so the link is known only to researchers. Our study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline with a completed checklist for observational cohort studies in epidemiology.

Study Population
We identified patients evaluated for NF1 or an NF1-related neoplasm from 1985 to 2020 (Figure 1; Table 1). An unblinded electronic health record review was performed by 2 research fellows (J.P.L.   8,9 Patients with incomplete data were excluded. Patient data were anonymized prior to analysis, and data abstraction was agreed upon prior to collection and reviewed during collection by the research team. In specific cases in which the 2 research fellows did not agree on the criteria of inclusion, a third blinded research expert (I.E.M.) was involved.
The primary outcome measure was disease-specific survival (DSS). Secondary outcome measures were comparisons of (1) overall survival of patients with NF1 with nonneurofibroma neoplasms vs those without nonneurofibroma neoplasms, (2) neoplasm prevalence in the NF1 group Surgical treatment was defined as surgical resection of the specific neoplasm, chemotherapy was defined as any systemic agent used to treat the specific neoplasm, and radiation therapy was defined as neoplasm-specific treatment using forms of radiation.

Statistical Analysis
Comparisons of neoplasm prevalence in the NF1 group with the general population were calculated using SEER prevalence data and large population databases referenced in Deaths from disease were considered a DSS end point; other deaths were considered censored observations. Causes of death were determined based on the documentation of the electronic health record. The causes of death were reviewed by the research team composed of physicians (J.P.L., Y.C., and K.E.T.). We calculated DSS using the Kaplan-Meier method; survival curves were compared using the log-rank test. We measured DSS from diagnosis date to date of neoplasm-specific death or censorship. P values were 2-sided, and P < .05 was considered statistically significant. Owing to the concern for type I error from multiple secondary analyses, we adjusted our significance threshold to P < .001 for secondary analyses. Statistical analyses were performed using SPSS statistical software

Results
Among 2427 patients evaluated for NF1 during the study period, 1607 patients were included ( Figure 1); 840 (52.3%) were female patients, and median (range) age at initial visit was 19 years (1 month to 83 years); 161 patients (10.0%) were younger than 8 years at last follow-up (        Among patients with NF1 without nonneurofibroma neoplasms, 12 patients presented with more serious multisystemic complications that were associated with significant morbidity and mortality. These patients died at a younger mean (SD) age (29 [8] years); causes of death included complications from facial or airway neurofibromas for 3 patients, deep neurofibroma resection for 3 patients, neurological deficits from neurofibromas for 3 patients, unknown causes for 2 patients, and pneumonia for 1 patient.

Glioma
Gliomas were the most common neoplasm in the NF1 group, found in 295 patients (18.4%). Within this subgroup, LGGs were the most frequently occurring, found in 267 patients (16.6%) (

Pheochromocytoma and Neuroendocrine Tumors
In the study group, 36 patients (2.4%) developed endocrine neoplasms ( (0.9-50.0) mm. Among patients with melanoma, 1 patient was diagnosed with periocular melanoma in the left orbit.

Discussion
Individuals with NF1 can develop a wide variety of neoplasms, and the overall risk of neoplasm development among individuals with NF1 is 5% to 15% higher than in the general population, with an earlier age of onset and worse prognosis. 24 This cohort study found that patients with NF1 had significantly lower DSS rates if they developed UPS, HGG, MPNST, ovarian carcinoma, or melanoma compared with other neoplasms. This study also found an increased prevalence of neoplasms among patients with NF1 (OR, 9.5; 95% CI, 8.5-10.5; P < .0001); however, this risk of neoplasm was primarily associated with tumors of the central nervous system and connective tissue. 6,24,25 To our knowledge, this study is the largest single-center cohort study to date evaluating the characteristics of neoplasia in patients with NF1.
The NF1 tumor suppressor gene encodes the neurofibromin protein, a negative regulator of the Ras oncogenic pathway. [26][27][28] Inactivating mutations in NF1 is associated with downstream activation of mitogen-activated protein kinase, phosphatidylinositol 3-kinase/protein kinase B/mechanistic target of rapamycin signaling, and uncontrolled cellular growth, differentiation, and survival, which is associated with the disease origin of NF1, including NF1-associated neoplasms. 2,[26][27][28] Few studies, to our knowledge, have examined the risk among individuals with NF1 of developing less common neoplasms or the outcomes associated with these neoplasms. 3  The most common LGG diagnosed in patients with NF1 was OPG, found in 178 patients (11.1%); however, this rate was lower than a 2007 review 29 estimate of 15% to 20%. This lower prevalence was likely associated with our observation that 551 patients (34.3%) in the NF1 group had no record of neuroimaging screening. Only half of children with NF1 and radiographically identifiable OPGs ultimately developed signs or symptoms of their tumor. Several non-OPG gliomas in our study group were detected and resected owing to symptoms. In addition, LGGs were significantly more frequent in our NF1 population than were sporadic LGGs in population estimates. A 2000 review 30 found that NF1-associated LGGs undergo malignant transformation more frequently than comparable sporadic LGGs; thus, malignant transformation of NF1-associated LGGs may be associated with the increased prevalence and younger age of diagnosis for HGG among patients with NF1 in our study compared with the general population. Despite aggressive multidisciplinary treatment, HGG had a significant association with mortality in our study group. These findings suggest that conservative management should be used in patients with NF1 and asymptomatic LGGs, treatment should be recommended among patients with symptomatic or progressing lesions, and HGGs should be treated aggressively with a multidisciplinary approach.
Our study found that patients with NF1 were at significant risk of developing MPNSTs (OR, 9043; 95% CI, 7840-10 431; P < .0001) and that MPNST diagnosis occurred at a young age (mean, 33.88 years). 30 The most common malignant neoplasm and cause of death in the NF1 group was MPNST; our findings agree with those from a 2012 study 31 and a 2017 study 32 that found a similar 5-year DSS among patients with NF1-associated MPNST. Often, MPNST-related symptoms overlap with those of growing neurofibromas, making diagnosis challenging. Larger tumors have been found to be indicators associated with a negative prognosis, suggesting the importance of early diagnosis and intervention for MPNST. 32 These findings suggest that frequent evaluation should be performed among patients with NF1, especially those who have significant pain or rapid change in size of an existing neurofibroma. 33 Although MPNST was the predominant sarcoma associated with NF1 in our study, non-neurogenic sarcomas were also more prevalent in the NF1 group than in the general population.

JAMA Network Open | Oncology
Cancer Prevalence in Patients With Neurofibromatosis Type 1 vs General Population In contrast to sporadic GISTs, of which most are found in the stomach, 65.0% of GISTs in our study group were located in the jejunum or ileum and 30.0% were found in the duodenum, consistent with the results of a 2005 study 34 that found that these tumors were associated with the small bowel.
Currently, there is no standard screening for GISTs; therefore, these findings suggest that careful attention should be given to patients with NF1 and possible GIST-related symptoms, such as recurring gastrointestinal pain, nausea and vomiting, positive fecal occult blood testing, or anemia. 35 Another non-neurogenic sarcoma prevalent in our study group was ERMS. In contrast to sporadic ERMSs, most NF1-associated ERMSs in our study occurred in the genitourinary system (84.6%) and there was a higher proportion of cases in male patients (male to female ratio of 3:1 vs 1.5:1 in sporadic cases). 23 In patients with NF1-associated ERMSs, the condition usually presents at an early age. Focused clinical exams with attention to common first symptoms (eg, scrotal mass, hematuria, or urinary retention) may be justified in boys with NF1. 10 Studies from 1990 to 2013 [36][37][38][39] found an association between NF1 and rarer sarcomas in our study, but these studies are limited to case reports. In our study group, NF1-associated UPSs were aggressive despite multimodal therapy; all patients diagnosed with UPS died from their disease.
These findings suggest that patients with NF1 should be frequently evaluated for symptomatic masses, given that all our patients with NF1 and UPS presented with a growing, painful mass.
Although osteosarcomas are also rare in patients with NF1, 2 studies 38,39 have found increased prevalence among these patients. Two patients in our study group had no evidence of osteosarcoma progression after therapy but ultimately died of a synchronous primary MPNST. Although rare, the association of leiomyosarcoma, liposarcoma, and angiosarcoma with NF1 may not be coincidental.
Larger series are needed to further elucidate the association. 38,[40][41][42][43][44] A 2010 study 1 found that individuals with NF1 have an increased risk of breast cancer. Our study results also found that young female patients with NF1 develop breast cancers frequently (OR, 3.8; 95% CI, 2.9-51; P < .0001), given that two-thirds of the breast cancers in our study group were diagnosed in women younger than age 50 years. These data suggest that recommendations should be followed from the National Comprehensive Cancer Network that screening mammography begin at age 30 in women with NF1. 45 In addition, 2 patients in our study group were diagnosed with breast cancer in their 20s, which suggests the importance of annual clinical breast exams after age 18 years and consideration of magnetic resonance imaging for symptoms or palpable breast masses.
To our knowledge, no study to date has evaluated the risk and association of ovarian cancer in patients with NF1. Our study found an increased prevalence and decreased age at diagnosis of ovarian carcinoma among patients with NF1 compared with the general population. Most patients with NF1 and ovarian cancer had developed peritoneal and distant metastasis by last follow-up, which may be associated with the high mortality rate observed. Analysis of larger series is necessary to distinguish whether ovarian cancer screening would be associated with benefits among women with NF1.
An association between pheochromocytoma and NETs has been observed in patients with NF1; indeed, because of this association, a diagnosis of 1 condition should prompt an evaluation for the other. 46 In our study, 1 patient with NF1 was diagnosed with both a pheochromocytoma and a rectal NET. Our findings suggest that these tumors should be considered in any patient with NF1 and patients experience disease recurrence despite successful resection. 49 Thus, long-term monitoring is indicated for all patients with NF1, even after successful resection. 49 Two studies 50,51 found an association between NF1 and melanoma, and our study found a significantly increased risk for melanoma among patients with NF1 (OR, 3.9; 95% CI, 2.4-6.5; P < .0001). Of our patients with NF1 and melanoma, 40.0% presented with metastatic disease at diagnosis, which is a higher proportion than in the general population (ie, 4%). 52 Furthermore, survival was worse in the NF1 group compared with the general population. Melanomas associated with NF1 melanomas had a median thickness of 2.7 mm, which supports previous findings that NF1-associated melanomas are thicker than sporadic melanomas (median thickness, 1.5 mm). 51 Our findings suggest the importance of maintaining a high index of suspicion for melanoma in patients with NF1 and performing frequent, meticulous skin and ocular exams.

Limitations
This cohort study has some limitations. First, some patients with NF1 may have received treatment elsewhere, although this would suggest a higher prevalence of neoplasia than that found in our analysis. Second, the patients lost to follow-up may be associated with changes in the outcome analyses. Third, 10.0% of our NF1 group was younger than age 8 years old at last follow-up, which may be associated with changes in rates of clinical characteristics and prevalence or risk of neoplasia in our group; these younger patients with NF1 may be associated with decreases in the true numbers of characteristics and patients at risk for non-NF neoplasia. Fourth, the mean age for patients with NF1 who did not develop additional neoplasms was similar to the mean age for those who did develop neoplasms. This may be associated with the small sample size. Despite these limitations, our study found neoplasia occurrence and patient outcome characteristics that may be used in the evaluation of neoplasia among patients with NF1.

Conclusions
This cohort study of 1607 patients found that patients with NF1 had significantly lower DSS rates if they developed UPS, HGG, MPNST, ovarian carcinoma, or melanoma compared with other neoplasms. Individuals with NF1 were found to have significantly increased rates of multiple neoplasms other than neurofibromas. Some of these are classically associated with NF1, and others were previously not associated with NF1. These neoplasms had significant associations patient outcomes. As larger studies of NF1associated cancers emerge, understanding of the neoplasia drivers and mechanisms associated with outcomes among individuals with NF1 may improve. This study's results may inform counseling of patients with NF1 and support a multidisciplinary approach to care.