Mortality associated with neurofibromatosis type 1: A study based on Italian death certificates (1995-2006)

Neurofibromatosis type 1 (NF1) is an autosomal dominant disorder which affects 1 in 2500-3000 live births [1, 2]. It is a progressive multisystem disorder characterized by a wide variety of clinical signs and symptoms, a totally unpredictable evolution, and an increased risk of malignancy. The gene responsible for NF1 encodes neurofibromin, a protein which acts as a tumor suppressor, the loss of which leads to an increased risk of developing tumors. The hallmark feature of NF1 is benign neurofibromas (a peripheral nerve sheath tumor); the other NF1-defining manifestations consist of cafè au lait patches, skin-fold freckling, iris Lisch nodules, optic pathway glioma, and bony dysplasia [3]. NF1 can also result in serious complications which affect diverse body systems and which are responsible for the deaths related to NF1; these include disfigurement, evolving scoliosis, cognitive or neurological impairment, vasculopathy, and malignancy, in particular, malignant tumors of peripheral and central nerve tissue [4‐11].

Although persons with NF1 have a decreased survival, information on NF1-associated mortality is limited. Only one nationally representative population-based study on NF1 mortality based on death certificates has been published [12], and the information on the causes of death among persons with NF1 mainly derives from case series. Moreover, given that NF1 has a variety of clinical manifestations and complications of varying degrees of severity, it may not be recorded as the official cause of death. An additional problem is that no specific code exists for NF1 in the International Classification of Diseases (ICD), so that it is not possible to distinguish it from other types of neurofibromatosis, such as neurofibromatosis type 2 (NF2). The importance of this distinction lies in the fact that NF1 and NF2 are clinically and genetically different, generally with different symptoms and requiring distinct medical management; nonetheless, they are often confused for one another by clinicians. Although NF2 is more aggressive, NF1 is numerically more important, in that its incidence is about 10 times greater than that of NF2 [13].

The objective of this study was to estimate NF1-associated mortality in Italy, in particular, the number of NF1-associated deaths in the past 12 years; we also compared the distribution of age at death to that of the general population, so as to have an indication of survival, and explored the relation between NF1 and other medical conditions.

Of the nearly 6 753 000 deaths in Italy in 1995-2006, 632 had a diagnosis of NF1 on the death certificate, yet for nearly three-fourths of them the underlying cause was coded as another pathology.

With regard to age, the bimodal trend in NF1-associated mortality emphasizes the weight of premature mortality (i.e., before 40 years of age): mortality dramatically increased from adolescence to 40 years of age, with one third (n = 185) of the deaths occurring before 40 years; after 40 years, the mortality decreased until 50 years of age and then showed a constant increase for older ages, with the highest absolute number of deaths among 70-74 year olds.

Compared to the general population, the mean age of NF1-associated deaths was approximately 20 years lower, and it was approximately 10 years lower when the analysis was restricted to persons 40 years of age and older, again demonstrating how the effect of NF1-associated mortality was mainly restricted to younger individuals, with the difference in survival between individuals with and without NF1 being mostly due to premature mortality from NF1 or NF1-associated causes. However, the finding that there was still a gap of 10 years when considering persons who died at 40 years of age or older suggests that NF1-associated disorders increase the risk of premature mortality also later in life.

As expected, among males, mortality was higher, though not significantly, and the mean age at death was significantly lower. However, the gender difference in age at death was smaller among persons with NF1 (4.4 years) compared to the general population (6.7 years), suggesting that women are affected by more severe NF1-related complications or that NF1 is more likely to be reported on the death certificates of young women. Some studies have shown that women with NF1 have a higher risk of malignancies than males [17, 18], though others have not [2, 19, 20]. However, during pregnancy and puberty, the progression to malignancy and the number and size of neurofibromas may increase in women with NF1 [21], and the greater attention placed on the health of women during pregnancy or childbearing could be reflected in a greater tendency for NF1 to be reported on the death certificates of young women.

The comparison of the two calendar periods may suggest that the survival of persons with NF1 has slightly increased, though longer time series need to be evaluated.

For the leading causes of NF1-associated deaths, we found an excess for malignant neoplasms of connective and other soft tissue and brain, which is not surprising. In fact, NF1 is unlikely to cause death, which is usually caused by NF1-related complications, mainly tumors of the peripheral or central nerve tissue [4‐6, 9, 11, 22‐28]. In particular, optic gliomas and cutaneous, subcutaneous and plexiform neurofibromas, which are diagnostic of NF1, can become malignant or they can remain benign yet cause pressure on the spinal cord or stenosis and occlusion of cerebral arteries. We also found a significant excess, though smaller or for very few cases, for neoplasm of unspecified parts of the nervous system, neoplasm of ill-defined/unspecified sites, and neoplasm of unspecified nature; when we reviewed these death certificates, we found that all of them mentioned involvement of the brain or nervous system; it is thus likely that the PMRs for malignant neoplasm of brain and connective and soft tissue were underestimated. Moreover, of the 150 deaths with neurofibromatosis recorded as the underlying cause, 17 mentioned lethal complications, and although all of these complications consisted of neoplasm of brain and connective soft tissue, they were not listed as the underlying cause. The decision as to what should be listed as the underlying cause could reflect differences in how individual physicians interpret the hierarchy of causes of death.

The estimated PMRs stratified by age confirmed the weight of premature mortality in NF1-associated deaths due to cancer.

With regard to non-neoplastic diseases, the 2-fold excess for obstructive chronic bronchitis among persons 40 years of age and older is consistent with the well-known respiratory manifestations of NF1. In NF1, the thorax and lungs can be affected in several ways, given the dense distribution of peripheral nerves throughout the thorax: cutaneous and subcutaneous neurofibromas on the chest wall and chest wall deformities; kyphoscoliosis; ribbon deformity of the ribs; thoracic neoplasms; parenchymal neurogenetic tumors [29‐31]; and other respiratory manifestations that severely impair pulmonary function, such as upper airway obstruction by neurofibromas, central hypoventilation, diaphragm paralysis, diffuse and interstitial lung disease, and primary pulmonary hypertension [32‐35].

The findings of the present study are consistent with those of the only other nationally representative population-based study on NF1 mortality using death certificates, which was conducted by Rasmussen et al. [12], as well as those of smaller survival studies based on the clinical follow-up of persons with NF1, performed in other industrialized countries [36]. In particular, the study by Rasmussen et al. [12], whose methodology was very similar to ours, found a mean annual frequency of NF1-associated deaths of 1/8 700 deaths for the period 1983-1997, which is slightly higher than our estimate of 1/10 685 deaths for a later period, 1995-2006. However, this difference could reflect temporal changes in the mortality of the general population rather than true differences in the survival of the NF1-population. In fact, Rasmussen found a mean age at death of White persons with NF1 of 55.4 years, which is very similar to our estimate of 55.5 years. Furthermore, in both studies, the difference in age at death between men and women was smaller among persons with NF1 compared to the general population. Moreover, the two studies show similar patterns of excesses/deficits for all conditions likely to be an underlying cause (i.e., malignant neoplasms of connective and soft tissue and brain). By contrast, the results differed with regard to other conditions (i.e., curvature of spine, epilepsy and mental retardation), which could be because Rasmussen analyzed all diagnoses reported on the death certificate, whereas we analyzed only the underlying cause.

Some potential limitations of this study need to be considered. Regarding the exclusion of NF1-associated deaths, we are quite confident that any bias would be negligible because the diagnoses were written out (as opposed to being coded), which allowed us to select nearly all death certificates with a diagnosis of NF1 and distinguish it from NF2. We searched for the terms "neurofibromatosis", "Von Recklinghausen disease", and specific symptoms, including synonyms, acronyms and short forms. Thus the only death certificates potentially excluded were those with a diagnosis entered so incorrectly that it eluded the dictionary of medical terms used by ISTAT. Regarding the erroneous inclusion of NF2 cases, as mentioned, NF1 and NF2 can be distinguished because the diagnoses are written out; they can also be distinguished when a diagnosis of neurofibromatosis with no other specification is reported yet the diagnoses of some complications/symptoms specific to NF1 or NF2 are reported. Although it is possible that the 215 cases with a diagnosis of neurofibromatosis without either specification or specific complications/symptoms included some cases of NF2, the mean age at death of these cases did not significantly differ from that of the other 418 cases with NF1, whereas the 23 cases excluded because of an evident or probable diagnosis of NF2 were significantly younger than the selected cases, which is consistent with the greater severity for NF2. Multiple-Causes-of-Death records were missing for the provinces of Trento and Bolzano, yet given that only 3 persons with neurofibromatosis were identified based on the National Mortality Database, this probably did not affect our results. Finally, the quality of data must be considered. Although the automated selection of the underlying cause of death has reduced coding and processing errors [37], the completeness and accuracy of the death certificate and the decedent's medical diagnosis remain as potential sources of error [38, 39]. Because physicians report only those conditions and co-morbidities that they believe (or know) to have been instrumental in causing death, it is likely that studies based on death certificates detect only severe NF1-cases, resulting in a potential underestimate of the life expectancy of persons with NF1. It is also likely that only associations with conditions well-known to be associated with NF1 were revealed, resulting in a potentially biased pattern of excesses/deficits for other clinical conditions. Furthermore, the PMR is probably underestimated for conditions that are frequent among NF1 patients and known to be associated with the disease, such as breast cancer, nervous system diseases, and hypertensive disease. This is because these diseases are unlikely to be considered as the underlying cause or to be directly involved in causing death when more aggressive complications are present, and in our study, to study the association between NF1 and other conditions, we were able to consider only the underlying cause (the only available coded cause reported on death certificates).

Finally, death certificates do not allow all persons with NF1 to be identified; thus the association with other conditions should not be interpreted as the risk of developing that condition among living persons with NF1.

Despite the limits discussed above, this study has important strengths. To the best of our knowledge, it is the first nationally representative population-based study on NF1-associated mortality in Italy. The combined use of the two national data sources allows for comparison with data from the general population and reveals the usefulness of already available data when a surveillance system is not fully operational. This study also stresses the importance of the Multiple-Causes-of-Death Database in providing a more complete picture of mortality for conditions that are frequently not recorded as the underlying cause of death, or to study complex chronic diseases or diseases that have no specific ICD code, such as NF1.

No funding was provided for this study.