Introduction

Pleuropulmonary blastoma (PPB), an embryonal sarcoma and the most common pediatric primary lung malignancy,1 is the hallmark tumor of the DICER1 syndrome.2 Somatic second hits in one of five “hotspot” amino acids of the DICER1 RNase IIIb domain are necessary for malignant transformation of PPB and other DICER1-associated tumors.3,4 Germ-line DICER1 mutations are also associated with a variety of other neoplasms, including cystic nephroma, nasal chondromesenchymal hamartoma, ciliary body medulloepithelioma, pituitary blastoma, as well as Sertoli-Leydig cell tumor and other ovarian sex cord stromal tumors.5

PPB is a serious complication of the DICER1 syndrome. PPB progresses through stages of malignant transformation, from type I (cystic) to type II (partially cystic/solid) and type III (solid). The 5-year overall survival for later-stage patients is 71% (type II) and 53% (type III). Relapse and metastasis, primarily to the central nervous system, is responsible for much of the mortality.6,7 Fortunately, early detection and subsequent surgical resection of PPB can be curative.8 However, this is feasible only if DICER1 mutation carriers are identified and screened (by chest computed tomography) as early in life as possible.

The presence of easily detected, nonmalignant phenotypic features can prompt alert clinicians to consider the diagnosis of an occult tumor-predisposition disorder in an otherwise healthy child. For example, the distinctive skin findings and increased head circumference in neurofibromatosis type 1 and Cowden syndrome may appear before any syndrome-associated neoplasia. In the DICER1 syndrome, lung cysts, cystic nephroma, or family history of multinodular goiter have been used to identify DICER1 mutation carriers (hereafter, “DICER1 carriers”) before the development of malignancy.8,9 Because the DICER1 syndrome was recognized relatively recently, systematic evaluations of growth have not been reported, although there have been accounts of developmental delay, macrocephaly, and overgrowth in patients with mosaic DICER1 “hotspot” RNase IIIb mutations.10 Other accounts have not reported overgrowth or developmental delay in these patients.11 In our natural history study of the DICER1 syndrome, we comprehensively evaluated individuals with germ-line DICER1 mutations and family controls. We analyzed auxology data from our cohort, particularly head circumference and height measurements, to characterize previously unrecognized DICER1-associated disease features that may be useful in identifying individuals and families at risk of PPB.

Materials and Methods

Study participants

The National Cancer Institute protocol “DICER1-Related Pleuropulmonary Blastoma Cancer Predisposition Syndrome: A Natural History Study” (National Cancer Institute protocol 11-C-0034; NCT 01257597) is open to individuals with DICER1-associated tumors and their family members. Between November 2011 and December 2014, 134 participants from 31 families were evaluated at the National Institutes of Health Clinical Center. CLIA-certified germ-line DICER1 mutation testing was conducted at Ambry Genetics (Aliso Viejo, CA) and the Children’s National Medical Center (Washington, DC). Five children without detectable germ-line DICER1 mutations but who harbored DICER1-associated tumors were considered separately. We compared individuals carrying pathogenic germ-line DICER1 mutations (n = 76) with unaffected family members lacking pathogenic DICER1 mutations (n = 53). The study was approved by the National Cancer Institute’s institutional review board, and all participants, or their parents or guardians, provided written, informed consent.

Dysmorphology and reference curves

Height was measured by clinical staff at the National Institutes of Health Clinical Center using stadiometers. Head occipitofrontal circumference (OFC), arm span, and lower-segment length were recorded using measuring tapes, as described by Gripp et al.12 Upper-segment length was calculated by subtracting the lower-segment length (distance from the top of the pubic symphysis to the floor) from the height. The ratio of upper-segment length to lower-segment length (US/LS) was then calculated. We compared observed measurements with age- and gender-appropriate reference charts for height13 and OFC.14 OFC and height are strongly correlated, and we used OFC-for-height references for those aged 18 years and older.15 We considered abnormal height to be below the 3rd centile or above the 97th centile and macrocephaly as greater than the 97th centile in published reference populations.

Statistical analyses

We tested differences in proportions using either a chi-squared test or an exact test when frequencies were low (n < 5). We assessed differences in continuous descriptive characteristics using the Wilcoxon rank-sum test. We fit generalized estimating equations to the continuous measurements of OFC and height in adults to account for correlation within families, and we used robust standard errors. All tests were two-sided; we considered P < 0.05 significant, and analyses were performed using Stata/SE version 13.1 (Stata, College Station, TX).

Results

Individuals (n = 19) were excluded from our analyses because of missing data on any of the following: OFC (n = 14), US/LS ratio (n = 11), and arm span (n = 9). The remaining 110 participants comprised the analytic data set. Cohort demographics are described in Supplementary Table S1 online.

Macrocephaly was more frequent in DICER1 carriers than in family controls

DICER1 carriers differed from family controls in the distribution of head circumference ( Figure 1a ): 28 DICER1 carriers (42%) were macrocephalic, and none had an OFC below the third centile, versus family controls, of whom five were macrocephalic (12%) and two had an OFC below the third centile (5%) (P < 0.001). This difference between DICER1 carriers and controls remained significant after stratification by gender. Seventeen females with a DICER1 mutation (50%) were macrocephalic, and none had an OFC below the third centile, versus three controls with macrocephaly (20%) and two with OFC below the third centile (13%) (P = 0.024). Similarly, 11 males with a DICER1 mutation (33%) were macrocephalic compared with only two controls (7%) (P = 0.026). When plotted against the reference curves published by Rollins et al.,14 OFC was systematically increased in DICER1 carriers compared with both population norms and family controls ( Figure 1b , c ). Among those aged ≥18 years, DICER1 carriers had a 2.25-cm increase in OFC (95% confidence interval: 1.2–3.3; P < 0.001), after adjusting for gender. We did not estimate the magnitude of the increase in children because the small number of pediatric controls precluded modeling of the nonlinear relationship between age and OFC. However, 8 DICER1 carriers aged <18 years (25%) were macrocephalic compared with one control aged <18 years (10%), though the difference was not statistically significant (P = 0.219). The five children with a DICER1-associated tumor but no detectable DICER1 germ-line mutation had a distribution of OFC similar to that of family controls ( Figure 1b , c ). OFC did not correlate with DICER1 mutation location or type (Supplementary Figure S1 online and Supplementary Table S2 online).

Figure 1
figure 1

Occipitofrontal circumference (OFC)–for–age in the DICER1 syndrome. (a) Abnormal OFC for age. Proportions between the 3rd and 97th centiles (white), below the 3rd centile (gray), or above the 97th centile (black) in DICER1 mutation carriers and family controls. P values are for the Fisher exact test of differences between groups. (b) Females. Red triangles indicate DICER1 mutation carriers. White diamonds represent family controls. Blue circles represent girls without a detectable germ-line DICER1 mutation but who harbor a DICER1-associated tumor (7-year-old: type II pleuropulmonary blastoma (PPB); 15.5-year-old: Sertoli-Leydig cell tumor; 17-year-old: type II PPB). The dashed lines indicate the 97th, 50th, and 3rd centiles of OFC for age reported in Rollins et al.14 The vertical dashed line at age 10 years indicates a change in the scale of the x axis to allow for better resolution of children’s values. (c) Males. Blue circles represent boys without a detectable germ-line DICER1 mutation but who harbor at least one DICER1-associated tumor (4-year-old: type I PPB and cystic nephroma; 7.7-year-old: type II PPB only).

Full size image

There were no differences in height between DICER1 carriers and controls

Although DICER1 mutation carriers were taller than familial controls after controlling for gender (P = 0.048), the proportion of individuals with height >97th general population centile was similar between DICER1 carriers (12%) and controls (7%) (P = 0.52) (data not shown). Tall stature was not more prevalent in either females (P = 0.414) or males (P = 1.0) with the DICER1 syndrome. The DICER1 syndrome in adults was not associated with greater height after adjusting for gender (difference = 2.5 cm; 95% confidence interval: −1.0–6.1; P = 0.160).

Larger head circumference in DICER1 carriers was independent from differences in height

As noted above, OFC and height are strongly correlated among the general population. Using the reference curves from Bushby et al.15 that adjust for height among those aged ≥16 years, the distribution of OFC for height among adults (age ≥18 years) with DICER1 differed significantly from that of family controls ( Figure 2a ). Ten adults with the DICER1 syndrome (29%) had an OFC above the 97th centile and none had an OFC below the 3rd centile, versus two controls with an OFC above the 97th centile (6%) and two controls with an OFC below the 3rd centile (6%) (P = 0.017). Stratifying by gender did not detect significant differences between groups. Eight women with the DICER1 syndrome (33%) had an OFC above the 97th centile and none had an OFC below the 3rd centile compared with one woman control (9%) with an OFC above the 97th centile and two (18%) with an OFC below the 3rd centile (P = 0.065). Two males with the DICER1 syndrome had an OFC above the 97th centile (18%) compared with one male control (5%) (P = 0.25). Qualitatively, OFC-for-height in DICER1 carriers was larger than expected ( Figure 2b , c ). Among adults, the DICER1 syndrome was associated with an average increase in OFC of 1.92 cm (95% confidence interval: 1.1–2.8; P < 0.001) after adjusting for gender and height.

Figure 2
figure 2

Occipitofrontal circumference (OFC)–for–height in the DICER1 syndrome (mutation carriers and controls ≥18 years old). (a) Abnormal head circumference for height. Proportions between the 3rd and 97th centiles (white), below the 3rd centile (gray), or above the 97th centile (black) in DICER1 mutation carriers and family controls. P values are for the Fisher exact test of differences between groups. (b) Females. Red triangles indicate DICER1 mutation carriers. White diamonds represent family controls. The dashed lines indicate the 97th, 50th, and 3rd centiles of OFC for height reported by Bushby et al.15. (c) Males.

Full size image

Arm span–to–height ratio and long-bone growth were proportional in patients with the DICER1 syndrome

No significant differences in the US/LS or arm span–to–height ratios were observed. The US/LS ratio in DICER1 carriers (mean = 0.977; SD = 0.087) was similar to that in family controls (mean = 0.950; SD = 0.087) (P = 0.443) among those aged 18 years and older. The relationship between arm span and height were linear, as is expected in the general population, and no significant differences were observed between DICER1 carriers (mean = 1.022; SD = 0.043) and controls (mean = 1.024; SD = 0.028) (P = 0.747). Adjustment for neither gender nor age meaningfully affected the results (data not shown).

Discussion

In our study macrocephaly was observed in 42% of DICER1 carriers evaluated at the National Institutes of Health Clinical Center. Other growth measurements were normal relative to the general population, that is, DICER1 carriers were not abnormally tall, but adults with DICER1 were taller on average compared with family controls. After adjusting for these differences in height, the association between macrocephaly and DICER1 mutation status persisted. Measurements of long-bone growth (arm span–to–height and US/LS ratios) were within normal ranges.

In mice Dicer1 is a haploinsufficient tumor-suppressor gene16; our data show for the first time that human macrocephaly is a phenotype significantly associated with DICER1 haploinsufficiency. In the DICER1 syndrome the macrocephaly is relatively (but not disproportionately) increased and is not associated with somatic overgrowth.17 Klein et al.10 reported macrocephaly and symmetric overgrowth in two children with mosaic missense “hotspot” mutations in the RNase IIIb domain of DICER1, along with developmental delay and Wilms tumor, in a constellation of findings they termed the “GLOW” (global developmental delay, lung cysts, overgrowth, and Wilms tumor) syndrome. The authors also identified 10 candidate dysregulated 3p microRNAs that target negative regulators of the mammalian target of rapamycin, transforming growth factor-β and mitogen-activated protein kinase signaling pathways, including PTEN, TSC, and NF1. They hypothesize that an imbalance in specific 3p microRNAs arising from DICER1 RNase IIIb mutations lead to excessive cell and tissue growth and tumor predisposition. Mosaic DICER1 RNase IIIb domain mutations are associated with a more severe neoplastic phenotype.11,18 Many of the GLOW phenotype features, including macrocephaly and overgrowth, were not observed in a set of four patients with mosaic DICER1 RNase IIIb mutations.11 The differences in these studies may be attributable to the pleiotropy and phenotypic variability inherent in mosaicism and highlight the need for a systematic, statistically grounded approach to syndrome delineation.

It is well known that haploinsufficiency of PTEN (Cowden and Bannayan-Riley-Ruvalcaba syndromes) and NF1 (neurofibromatosis type 1) is associated with macrocephaly. It is interesting to note that these genes are also dysregulated in DICER1 mosaicism.10 Increased head circumference in neurofibromatosis type 1 is hypothesized to be a secondary skeletal manifestation of brain overgrowth,19 presumably caused by dysregulation of key growth pathways. The often pronounced macrocephaly in these disorders can be a useful clinical clue to their diagnosis. The role of these genes as intermediaries of posited brain overgrowth with secondary skeletal growth in the DICER1 syndrome phenotype merits further study.

Our analysis is limited by the biases inherent in using cross-sectional data to assess growth. Longitudinal analyses are needed to discern when OFC increases and would inform future studies of the underlying mechanism of this growth. Moreover, families enrolled in the study were accessioned because of a history of a DICER1-associated tumor. Ascertainment bias may have missed clinically asymptomatic DICER1 carriers with milder phenotypes. Finally, measurements were made by multiple observers rather than a single physician. However, the measurements of OFC, arm span, and height are unlikely to vary substantially enough between observers to account for the large difference observed in DICER1 carriers in this study.

In summary, our study is the first to document macrocephaly as a non-neoplastic feature of the DICER1 syndrome. Further analyses of longitudinal data may shed light on the developmental processes underlying the macrocephaly and point to the role of DICER1 in auxology. Like other, better-known tumor-predisposition disorders, macrocephaly may be a useful, if subtle, clinical clue to the diagnoses of the DICER1 syndrome.

Disclosure

The authors declare no conflict of interest.