American J o u r n a l of Medical Genetics 35:105-114 (1990)

Facial Morphometrics in the Identification of Gene Carriers of X-Linked Hypohidrotic Ectodermal Dy splasia Sudha Srikrishen Saksena and David Bixler Department of Oral-Facial Genetics, I.U. Medical Center, Indianapolis, Indiana

Roentgenographic measurements and morphometric analysis were employed in the investigation of contrasting patterns of craniofacial variation between normal individuals and those affected by X-linked hypohidrotic ectodermal dysplasia (HED). The research objective was to identify and describe the facial characteristics of heterozygous gene carriers who show minor expression of the disorder. In this study of 13 HED families with 16 affected males, 12 carriers, and 12 normal individuals, affected individuals had at least 3 of the following4 clinical signs and symptoms: a) hypodontia, b) hypohidrosis, c) hypotrichosis, and d) clinically distinct facial physiognomy. By contrast, the gene carriers manifested only one or 2 or none of the 4 clinical manifestations. In a preliminary comparison of gene carriers vs. normal individuals, we have generated 2 discriminant functions (each based on 3 facial measurements taken either from the lateral or frontal cephalograms). These 2 functions correctly diagnose 100%of the gene carriers and normal HED relatives. Facial anomalies characteristic of the gene carriers were 1)abnormally narrow and short maxillary width and palatal depth dimensions; 2) very small and retrusive malar and maxillary regions; 3)markedly reduced lower facial depth, height and width dimensions; 4) small head height, prominent forehead, and high-set orbits; 5 ) a generalized, symmetric reduction of the whole craniofacial complex.

Received for publication March 21, 1989; revision received July 26, 1989. Address reprint requests to Dr. Sudha S. Saksena, Department of Oral-Facial Genetics, I.U. Medical Center, 1226 West Michigan Street, BR 026, Indianapolis, IN 46202.

0 1990 Wiley-Liss, Inc.

KEY WORDS: clinical morphometry, craniofacial pattern recognition (CFPP), roentgencephalometry, discriminant analysis, morphometric diagnosis, heterozygote detection, X-linked hypohidrotic ectodermal dysplasia.

INTRODUCTION Patients with hypohidrotic ectodermal dysplasia (HED) are characterized by the clinical manifestations of hypodontia, hypohidrosis, hypotrichosis, and a highly characteristic facial physiognomy. This disorder is inherited as a n X-linked trait, although some authors support the existence of a n autosomal recessive form a s well. Affected males typically show complete expression of clinical findings, whereas affected females representing lyonization effects exhibit a wider spectrum of signs, ranging from very subtle to the most severe expression of the disorder. Since the females a t the mild end of this spectrum are not always easily diagnosed as gene carriers, the prevalence of HED in females is difficult to estimate. The probability of detection of a mildly affected carrier female on the basis of a physical examination has been estimated to be about 70%, with detection estimatesranging from 42 to 88%[Settineri et al., 1976; Pinheiro and Freire-Maia, 1979; Airenne, 1981; FreireMaia and Pinheiro, 19841. Other methods which can be of aid in carrier identification are sweat-pore counts [Kerr et al., 1966; Frias and Smith, 1968; Kleinebrecht et al., 19811, hair line distribution patterns of Blaschko [Happle and Frosch, 19851, pigmentation patterns [Elejalde and Elejalde, 19831, and missing and/or misshapen teeth [Nakata et al., 1980; Sofaer, 1981a, 1981bI. More recent use of restriction fragment length polymorphisms (RFLPs) and DNA probes [MacDermot et al., 1986; Clarke et al., 1987; Zonana et al., 19881to localize and isolate the HED gene appear promising. Thus, it is clear that use of currently available clinical methods does not allow the routine identification of the HED heterozygous gene carriers.

Saksena and Bixler

106

Facial physiognomy of fully affected HED individuals in both sexes (from a wide variety of ethnic groups and nationalities) is strikingly similar and highly diagnostic. Consequently, it is reasonable to propose that an objective delineation and identification of facial morphology of HED affected vs. normal individuals can be developed to allow the identification of the mildly affected gene carriers. The present study employed roentgenographic measurements and multivariate analysis for the characterization, classification, and identification of HED gene carriers. In particular this study 1) identifies and describes HED-specific facial anomalies, 2) develops discriminant function equations that efficiently distinguish affected from unaffected individuals, 3) tests the diagnostic efficacy of the derived discriminant functions for the correct identification of individuals who are shown by pedigree to be obligate gene carriers of the trait but who show very mild (if any)

clinical expression of the disorder, and 4) presents a morphometric method for diagnosing HED relatives with unknown or uncertain clinical status but who are a t risk for transmitting the abnormal HED gene to their offspring.

PATIENTS AND METHODS Sixty individuals from 13 HED families were studied. Pedigrees of these 13 families showing major clinical finding are given in Figure 1 A, B. Pedigrees #619, #23117, #23333, #26952, #53053, #82902, #92273, and #I81838 appear to represent an X-linked form of HED, while pedigrees #22052, #28476, #53054, #81748, and #81865 are compatible with either an X-linked or autosomal recessive inheritance. Phenotypically, the autosomal recessive form is reported to be indistinguishable from the X-linked type [Goodman and Gorlin, 19831. In our sample, affected individuals

A FAMILY + 6 1 9

FAMILY ,231 1 7

FAMILY +23333

FAMILY

+

26852

I

I

:z

FAMILY + a 2 9 0 2

FAMILY +53053

1;2

3

-

FAMILY + 8 2 2 7 3

?-r@

4

B FAMILY ,22052

FAMILY + 2 8 4 ? 6

FAMILY + a 1 7 4 8

FAMILY +81838

FAMILY + 5 3 0 5 4

FAMILY + e l 8 6 5

I

Fig. l . A Pedigrees of families with X-linked hypohidrotic ectodermal displasia. 0, abnormal teeth (missing,misshapen); Q, decreased sweating; sparse, fine hair; 0, distinct face; *, cephalometric data. H: Pedigrees of families with either a n X-linked or a n autosomal recessive inheritance pattern of hypohidrotic ectodermal dysplasia. Q, abnormal teeth (missing, misshapen);@, decreased sweating; B, sparse, fine hair; 0,distinct face; *, cephalometric data.

e,

Facial Morphometry exhibited a t least 3 of the following 4 clinical signs and symptoms: a) hypodontia b) hypohidrosis, c) hypotrichosis, and d) a characteristic face. The gene carriers manifested only one or 2 of the clinical findings. With the foregoing criteria, 16 individuals (11 males and 5 females) were classified as affected, 12 individuals (all females) as gene carriers, and 12 individuals (10 males and 2 females) a s normal. This sample was thought to be diagnostically homogeneous. The remaining 20 individuals (11males and 9 females) who could not be clinically diagnosed were designated as unknown. Lateral view (LA) X-ray headplates were studied in all 60 individuals; however, only 52 frontal (PA) headplates (13 affected, 10 gene carriers, 10 normal and 19 unknowns) could be obtained for study. A total of 74 linear and angular measurements from both the LA and PA cephalograms were used to define the size and shape of the major anatomic areas of the head and face. Landmarks were digitized directly from the radiographs. In order to reduce landmark identification errors, each radiograph was digitized on 2 separate occasions, and the computer-generated measurements were compared and verified. If there was a discrepancy of one or more millimeters, the radiograph was digitized for a third time. Linear measurements were computed to the nearest 0.5 mm and angular measurements to the nearest 0.5”. Our study sample of 60 individuals included 32 males and 28 females, with 50% ofthe individuals ranging in age from 3% to 25 years. To adjust for age and sex variation in our sample, we tested using 2 sets of regression weights. To detect the interaction of both age and sex effects on measurements, a stepwise multiple regression analysis was conducted using each measurement as a dependent variable, and age and sex to a second degree polynomial (i.e., age, sex, age2,age x sex, and age2 x sex) as independent variables. One set (of regression weights) was generated from the present study sample of HED relatives (affected, unaffected, and unknowns) and a second set was generated from a normal population sample of 439 males and 537 females for the LA view measurements [Saksena et al., 19871 and 214 males and 213 females for the PA view variables [Saksena 19891. All were North Americans of European ancestry ranging in age from 3 % t o 25 years. In the discriminant analysis of the HED affected including gene carriers versus normal individuals, when regression weights generated from the total HED sample were employed, a 100%correct classification was attained for each group. On the other hand, when regression weights from the normal data base were used, affected and normal individuals were 100% correctly classified, but 4% to 8%of the gene carriers were misclassified as normal individuals. Thus, for making adjustments for age and sex effects the HED-specific regression weights were employed. No age and sex effects were found for the following PA and LA measurements: AP-CRO, APMOL, AP-MOR, CNS-SD, ID-ME, S-N-Apt., S-N-Bpt., Apt.-N-Bpt., and N-Apt.-PG. In 8% of the study sample, the vertex region was missing on the radiographs. Two statistical packages [SPSSX Inc., 1988; SAS Institute Inc., 19861 were used in the data analysis.

107

The following landmarks (anatomic and derived) and reference lines were employed for taking measurements: anterior nasal spine (AN); apex (AP);articulare (AR); basion (BA); ear-rod point (ER pt.); euryon (EUleft side, EU’-right side), the most lateral point at the parietal surface; frontotemporale (FT-left side, or FTL; FT’-right side, or FTR), the most medial point on the incurve of the temporal ridge; glabella (GL), the most anterior point between the supraorbital ridges; gonion (GO-left side, or GOL; G0’-right side, or GOR); infradentale (ID), the highest point on the lower alveolar margin between the central incisors; lateroorbitale (LOleft side, or LOL; LO‘-right side, or LOR), the lateral most point on the lateral orbital margin; midfrontal point (MF pt.), the most anterior and superior point on the frontal bone; mastoidale (MS-left side, or MSL; MS’right side, or MSR), the lowest point of the mastoid process; maxillary notch (MX-left side, or MXL; MX’right side, or MXR), the most medial point on the maxillo-alveolar surface; medioorbitale (MO-left side, or MOL; M0’-right side, or MOR), the most medial point on the medial orbital margin; menton (ME), the most inferior point on the midline border of the chin region; nasion (N), nasal cavity (NC-left side, or NCL; NC‘-right side, or NCR), the most lateral point in the nasal cavity; nasal shelf (NS-left side, or NSL; NS’-right side, or NSR), the lowest point of the outline of nasal shelf; opisthocranion (OP), the most posterior point on the occipital bone; orbitale (OR), the lowest point of the inferior border of orbital margin; pogonion (PG), the most anterior point of the chin region; posterior nasal spine (PN); pterygomaxillary fissure inferior (PT); rhinion (R); roof of orbit (RO-left side, or ROL; R0’-right side, or ROR), the most superior point on the roof of the orbit; sella ( S ) ;supradentale (SD), the lowest point on the upper alveolar margin between the central incisors; subspinale (A pt.); supramentale (B pt.); zygoma (ZYleft side, or ZYL; ZY’-right side, or ZYR), the most lateral point on the zygomatic arch; zygomaxillary inferior (ZI), the lowest point on the zygomatico-maxillary suture; zygomaxillary superior (ZS), the highest point on the zygomatico-maxillary suture; CRO-midpoint on the Ro to RO’ line (supraorbitale-left side, or ROL, to supraorbitale-rightside, or ROR); CNS-midpoint on NS to NS’ line; ZY to ZY’, or Z-line; and M S to MS’, or BS-line. In order to test landmark discernment errors, subsets of measurements were taken and compared from landmarks in close proximity, such as ZS and OR, ME and PG, PN and PT, AN and A point. Measurements such as ZS to ZI and OR to ZI, S to ME and S to PG, ERpt-PG, and ERpt-ME, GO to ME and GO to PG, S-NIZS-ZI and S-NI OR-ZI, etc., were found to be highly correlated. Of course? all paired (left and right) facial dimensions were also highly correlated [Saksena, 19891.

RESULTS Facial Morphology In order to identify and describe the HED-specific craniofacial measurements, a univariate one-way analysis of variance of gene carriers vs. normal relatives was used 1)to test the significance of variable mean differ-

108

Saksena and Bixler

ence and 2) to assess the relative contribution of each measurement to the 2 groups discrimination. Facial dimensions quantified were as follows: lA, facial height; l B , facial height and symmetry; 2A, facial width; 2B, facial depth and cranial base; 2C, facial height-widthdepth and symmetry; 3, facial profile angles. In Appendix A are cited age and sex-adjusted means, standard deviations, mean difference, and significance of univariate F value. Measurements most useful in the facial pattern recognition of the HED gene carriers were1. abnormally narrow and short maxillary width and palatal depth dimensions (MX-MX’, NS-MXL & NS’MXR, PN-AN) 2. very small malar bones, narrow forehead and midface width, depressed nasal bridge and highly retrusive maxilla (OR-ZI, RO-RO‘, ZY-ZY‘, ZS-N-R, GL-N-R, S-AN, S-PN, ER pt.-OR, S-N-Apt.); 3. very small lower facial depth, height and width dimensions (S-PG, S-GO, S-ME, ER pt.-ME, AR-PG, AR-GO, GO-PG, ME-GOR, ID/ZY-ZY’, GO-GO’); 4. small head height, prominent forehead and highset orbits (AP-CRO,APIMS-MS’,MFpt-GL-N, ROLIMSMS’ & RORIMS-MS‘); 5. low-set mastoids and narrow cranial base (ZYMSL & ZY-MSR, MS-MS’); 6. small cranial length and total facial height (GLOP, N-ME); and 7. a generalized, symmetric reduction of the face and head dimensions. In the past facial characteristics of the HED affected individuals have been described on the basis of clinical evaluations [Weech, 1929; Thannhauser, 1936; Singh et al., 1962; Reed et al., 1970; Brownstein, 1973; Gorlin et al., 1976; Soderholm and Kaitila, 19851 or quantitative facial findings [Brodie and Sarnat, 1942; Sarnat et al., 1953; Lipshutz, 1963; Harbour, 1981; Ward and Bixler, 1987; Bixler et al., 19881, as having a prominent forehead, a depressed nasal root and bridge, small cranial length and facial height, a n underdeveloped midface, a small palatal depth, and a short lower face with a small chin. Morphometric facial evaluation supported the presence of these same craniofacial anomalies in the HED gene carriers.

Discriminant Analysis Discriminant analysis (DA) was undertaken to identify sets of anatomical measurements (from the LA and PA view radiographs) that contribute most to group discrimination (gene carriers vs. normal). Discriminant analysis involves consideration of a number of measurements simultaneously in 2 or more groups. The important distinction to this approach of analyzing data is that the multiple measurements are considered in combination as systems. Univariate analysis in dealing with each dimension or measurement does not take into consideration the interrelationship of morphological characteristics, either functionally or statistically, nor does such a n analysis deal with the choice per se of their significance as characteristics in group discrimination. Thus, the results of such a comparison may be difficult to

interpret. A skull or a bone is a unit and i t should be treated as such in its total morphological pattern. According to Bronowski and Long [1952], “the right statistical method must treat the set of variates as a single coherent matrix.” Such is the technique of DA which expresses variation in the total morphological pattern in a single objective meaningful way. The discriminant function procedures LHowells, 1973; Oxnard, 1973; Hand, 1981; Brown, 19841 are capable of providing a) discrimination relating to multiple measurements taken on many individuals within 2 or more groups; b) a better understanding of the relationship among the different measurements taken on each individual; and c) diagnosis of individuals with either unknown or uncertain group status. DA is now widely available to researchers and clinicians due to its inclusion in packaged computer programs [such as SPSSX Inc., 1988; SAS Institute Inc., 1986; and BMDP Statistical Software, Inc. 19881. The relative ease of use of this method may open it to potential misuse when the basic assumptions of the method are not met. The important assumptions that must be examined are 1)the total number of cases (individuals) in each group should be several times greater than the number of discriminating variables used in the discriminant function; 2) each discriminating variable must have a normal distribution; and 3) they also must have equal variance-covariance matrices within each group. These assumptions were adhered to in the analysis of our data. Since the sample size of gene carriers and normal individuals is small compared to the number of variables per case, the data were subjected to a stepwise discriminant analysis. Thereby, a sequential selection of subsets of measurements, not more than 6 to 8 a t a time, was employed to test the relative importance of both LA and PA view measurements as discriminators.

DISCUSSION Several discriminant functions based on 3 or more facial measurements (taken either from the LA or PA radiographs) have been derived. These functions can discriminate 100% correctly the gene carriers from normal individuals. In Tables I & I1 are given the 2 allaround “best” discriminant functions: 1) Equation I based on 3 LA measurements S-PG, OR-ZI, and GL-OP and 2) Equation I1 based on 3 PA measurements ID!’ZYZY’, MX-MX‘, and ME-GOR, respectively. The standardized discriminant coefficients represent the relative contribution of each measurement in the 2 group discrimination. The unstandardized coefficients were used to obtain a discriminant score for each individual. The score for each individual was calculated by multiTABLE I. Discriminant Function Coefficients for Gene Carriers vs. Normal Individuals LA measurements

S-PG OR-ZI GL-OP (Constant)

Standardized coefficients

Unstandardized coefficients (Eq. I)

0.5909 0.5787 0.2769

0.1778 0.3561 0.1398 - 59.7557

Facial Morphometry

109

The girl is classified as affected. The mean and standard deviation of the discriminant scores for the gene carriers are -2.45 1.0, and for the normal relatives 2.45 I Standardized Unstandardized 1.O, respectively. Figure 2 shows a histogram ofthe score PA measurements coefficients Coefficients (Eq. 11) distribution. Discriminant function Equation I (based IDiZY -ZY' 0.4481 0.1438 on only 3 measurements: S-PG, OR-ZI, and GL-OP) sepMX-MX' 0.4483 0.3429 arated 100% of gene carriers from normal individuals. 0.2395 0.2052 ME-GOR A second function, Equation 11, given in Table I1 (Constant) 42.2020 (based on 3-PA view measurement IDIZY-ZY', MXMX', and ME-GOR) can also separate 100% of the gene plying raw values of each measurement by its corre- carriers from the normal relatives. The histogram in sponding unstandardized coefficient and adding the Figure 3 shows the distribution of discriminant scores product plus the constant. The following steps are in- based on the 3 PA view measurements. The means and volved in the application of discriminant function Equa- standard deviations of the discriminant scores for the tion I for morphometric identification of HED patients: gene carriers and the normal individuals are - 2.25 +1.0, respectively. The 2 discriminant 1.0 and 2.25 Step 1. Take the following three linear measurements function equations (I & 11)appears to be equally power(to the nearest 0.5 mm) on a lateral cephalometric ful in discriminating gene carriers from normal individradiograph. uals. However, for practical purposes further discussion a. Xi (S-PG) is focused only on the validation of discriminant function b. Xz (OR-ZI) Equation I. C. Xs (GL-OP) In Table I11 are listed the discriminant scores and Step 2. Make adjustments for age and sex variation for predicted group membership status for the 20 unknown each measurement (Y1, Yz, and Y3)in Step 1using the cases (11 males and 9 females ranging in age from 4 following HED data-specific regression equations. years to 55 years). The clinical status of these individ(Age is actual age in years and months in decimal uals is either uncertain or unknown. Pedigree informapoint; if age equal to or greater than 25 years then age tion for each case is given in Figure 1A,B. In this group, = 25; if actual age is less than 25 years, then the there was no history of parental consanguinity. All males = 1.1and females = 1.2; I1 = age x sex and12 mothers of probands U-2, U-5, U-9, U-12, and U-16, = Age' x sex.) designated as unknown because of lack of clinical signs, Y1 (adjusted) = X1 + 2.5722(25-Age) - 0.0261 were identified by this methodology as gene carriers. (625 - Age2) - 0.323(25 - 11) Similarly, all husbands of these mothers, fathers U-3, Y2 (adjusted) = X2 0.3397(25-Age) U-6, U-10, U-13, and U-17, were found to fit the normal - 0.0022(625 - 12) diagnosis. Of the remaining 3 females U-4, U-8, and Y3 (adjusted) = X3 + 0.9042(25 - Age) U-19, for which the pedigree and clinical exams were not - 0.3483(25 - 11) helpful, this method identified U-4 and U-8 as normal Step 3. Enter the three adjusted variable values (Y1, YP, and U-19 as a gene carrier. U-19, a young female, will be and Y3) into the followingdiscriminant function equation. Discriminant score = 0.1778Y1 + 0.3561Y2 + 0.1398Y3 - 59.7557 TABLE 11. Discriminant Function Coefficients for Gene Carriers vs. Normal Individuals

*

~

*

+

If the value of the discriminant score for a given individual is above zero, the individual is classified as normal. If below zero, the individual is affected (i.e., gene carrier). As an example, the discriminant score for a 4.3year-old girl is calculated as follows: Step 1. Her three linear measurements at 4.3 years agea. X1(S-PG) = 99.8 mm b. X2 (OR-ZI) = 12.8 mm c. Xs (GL-OP) = 184.3 mm Step 2. Her predicted adult measurements will beY1 = 99.8 + 2.5722(25-4.3) - 0.0261(625-4.32) - 0.332[25 - (4.3 x 1.211 = 131.20 mm Yz = 12.8 + 0.3397(25 - 4.3) - 0.0022[625 - (4.32 x 1.2)] = 18.64 mm Ys = 184.3 + 0.9042(25-4.3) - 0.3483[25-(4.3 x 1.2)] = 196.27 mm Step 3. Her discriminant score is(0.1778 x 131.20) + (0.3561 x 18.64) + (0.1398 x 196.27) - 59.7557 = -2.35

-6.0

-40

)20

0

20

'

4.0

6.0

Fig 2 Histogram of discriminant scores (based on 3 LA measurements) gene carriers, normal individuals

a,

n,

41

Fig 3 Histogram of discriminant scores (based on 3 PA measuregene carriers, 0, normal individuals ments)

a,

110

Saksena and Bixler TABLE 111. Subjects With Unknown or Uncertain (U) Clinical Status

ID No.

u-1 u-2 u-3 u-4 u-5 U-6 u-7 U-8 u-9 u-10

u-11 u-12 U-13 U-14 U-15 U-16 U-17 U-18 u-19 u-20

Pedigree No.”

Relationship to DroDositus

Age‘

Sex.

619-111-1 22052-1-2 22052-1-1 26952-111-1 28476-1-2 28476-1-1 28476-11-2 53053-11-1 53054-1-2 53054-1-1 53054-11-1 81748-1-2 81748-1-1 81748-11-1 81838-11-2 81865-1-2 81865-1-1 92273-11-3 92273-111-2 92273-111-3

Brother Mother Father Sister’s daughter Mother Father Half-brother Sister Mother Father Brother Mother Father Brother Mother Mother Father Mother’s brother Mother’s sister’s daughter Mother’s sister’s son

24.3 31.3 30.5 9.4 31.9 37.10 5.9 5.3 33.0 33.0 7.0 31.4 31.0 4.8 55.2 24.0 27.0 24.7 6.11 5.0

M F M F F M M F F M M F M M F F M M F M

Discriminant score 0.58 -3.36 0.59 0.83 -2.14 3.06 1.50 0.32 -2.80 0.26 1.65 - 0.53 5.72 2.04 - 2.29 1.30 1.62 0.56 - 2.56 0.93 -

Predicted group Status Normal Carrier Normal Normal Carrier Normal Normal Normal Carrier Normal Normal Carrier Normal Normal Carrier Carrier Normal Normal Carrier Normal

“See pedigrees in Figure 1. bAge in years and months. ‘M-male: F = female.

followed closely for signs of HED, since her mother and 2 maternal aunts are reported to be affected. Both young and old males (U-1, U-7, U-11, U-14, U-18, and U-20) were included in the unknown category for 2 reasons: a) to verify their classification and the identification accuracy of the function equation; and b) for 2 cases (U-1 and U-181, to consider the possibility that fine, blondish red hair in them was a familial trait and not syndromic. All 5 men were identified as unaffected. Thus, the presence of fine, blondish red head hair in the 2 males appears to be a familial or ethnic trait. The discriminant scores were 0.58 and 0.56 for U-1 and U-8, respectively. The discriminant scores for the mothers and fathers of HED children ranged from - 3.36 to - 0.56 and from 5.72 t o 0.26, respectively. Thus, in each family all mothers of affected individuals were identified as heterozygous gene carriers and all fathers were unaffected or normal. Therefore, by this evaluation none of the clinically normal mothers of HED affected offspring proved to be free of the gene. From this limited sample size, 2 important genetic conclusions may be drawn. First, the mutation rate for the HED gene appears to be very low. Second, since all the husbands of these gene carrier mothers (fathers of affected individuals) were designated as normal, this result gives no support for the existence of an autosomal recessive form of HED. If heterozygous females can be designated by this method, heterozygote males (autosomal) should also be identifiable. The lack of such identification suggests in this small sample that all the HED cases reported here are of the X-linked type and all have gene carrier mothers. Case U-15, the 55-year-old mother of both a fully affected male and a carrier female, had no apparent signs of HED herself and was by pedigree an obligate gene carrier, Her discriminant score (DS) of - 2.29 confirmed her carrier status. Her mother and mother’s sis-

ters reportedly had a “somewhat” reduced sweat-pore count, and one maternal aunt had a fully affected son. Case U-16, another asymptomatic mother of a fully affected female who is clinically normal also had a “somewhat” reduced ability to sweat. Her DS was -1.30, which classified her as a gene carrier. Case U-12, a normal mother of a fully affected male had a history of premature baldness in her family and a mother with serious eczema problems. Precocious baldness and eczema have been noted in “apparently healthy” but affected relatives of a large family with X-linked HED [Settineri et al., 19761.We classified case U-12 as a gene carrier; her DS was - 0.53. Case U-5 is another asymptomatic mother of a fully affected female with a history of 2 miscarriages (one with an ectopic pregnancy). Her DS was - 2.14 which classified her as a gene carrier. In the remaining 2 mothers, U-2 and U-9, both with fully affected sons and negative family history, the discriminant scores were - 3.36 and - 2.80. This identifies them as gene carriers. Their discriminant scores were actually above the mean value of - 2.45 which was obtained for the known-carriers. In order t o verify further the classification accuracy of discriminant function equation I, discriminant scores were calculated for 16 known-affected individuals (5 females and 11males). In the derivation of Equations I and I1 these affected individuals were excluded from the sample. The morphometric diagnosis identified 15 of the 16 clinically affected individuals correctly. A female (92273,II-6)had the highest discriminant score ( - 4.43) and a male from another family (26952,II-4) scored the second highest ( - 3.90). The discriminant scores for the affected males and females did not differ significantly. A known-affected9-year-old proband (23333-11-1)was misclassified as normal. Following a detailed examination of his craniofacial measurements, he was found to

Facial Morphometry

have both extreme n o w t h retardation of some specific craniofacial dimensFons and a dramatic over-growth of other parts. The bilateral width of his lower midface (NS-MXL and NS-MXR), maxillary width (MX-MX’), infradental height (IDIZY-ZY’),malar heights (OR-ZI, ZS-ZI), posterior facial height (S-GO), and cranial base width (MS-MS’) showed the greatest growth deficiency. These dimensions strongly confirmed his clinical HED diagnosis. However, unlike the other affected individuals, the minimum forehead width (RO-RO‘),interorbital width (MO-MO’),head height (AP-CRO), minimum nasal width (NS-NS’),mandibular depth (GO-ME, GOPG) and cranial length (GL-OP) were 2 SD above the mean. What syndromic or familial factors may be responsible for this altered growth pattern are not clear. However, one thing this result makes clear is that the selection of variable entered for discriminant function(s1 will influence the accuracy of identification and classification of individuals. Thus, when we used discriminant function Equation I1 based on 3 measurements from the PA headplates (see Table I1 and Appendix B), no misclassification occurred and all known-affected individuals were correctly identified. Discriminant function Equation I (based on 3 measurements from the LA radiographs) and Equation I1 (based on 3 PA measurements) provide a 100% correct classification and identification of affected vs. unaffected individuals (Figs. 2, 3). Of the 3 LA measurements, malar height, OR-ZI, and midfacial depth, S-PG, are of equal importance as discriminators; their respective standardized coefficients (SC) are 0.58 and 0.59 (Table I).The cranial length GL-OP with an SC of 0.28 is only half as good a group predictor. In the 3 PA measurement subset, maxillary width MX-MX’ and infradental height IDIZY-ZY’ are equally good as discriminators; the SC value of each measurement is 0.45. The third variable ME-GOR (mandibular width-right side) has a n SC of 0.24 and is about half as influential a discriminator. In the discriminant analysis of the HED relatives (affected, gene carrier, and normal individuals), it appears that both function Equations (I & 11) are about equally good in terms of classification and/or identification accuracy. However, the facial depth, height, and length measurements taken on a lateral radiograph are more diagnostic ofthe HED face (Appendix A). Thus, for clinical appraisal of individual cases the use of LA headplate alone is sufficient. Roentgencephalometry is a useful diagnostic tool, and when properly used it can be a valuable aid in the description and identification of HED affected individuals. Only a n elementary knowledge of the cephalometric radiograph is required to make a rapid and reliable morphometric diagnosis of the asymptomatic gene carriers and this method can be of considerable aid in the genetic screening of HED families. This is particularly so for identifying mildly affected X-linked female gene carriers who have a n uncertain clinical status but who appear to be at risk for transmitting the disorder. This method will now permit us to identify correctly individuals carrying the gene for HED and thereby provide accurate genetic counseling regarding recurrence risks for parents.

111

ACKNOWLEDGEMENTS Research for the senior author (S.S.S.)was supported by NIH-NIDR Research Career Development Award DE00126. Computer facilities were provided by IUPUI Computing Services. Thanks are extended to Mr. P. Kizer of the IUPUI Computer Center for his assistance in data analysis. We are grateful to Mary Kaye Richter, Executive Director, National Foundation for Ectodermal Dysplasia, for providing access to the HED patients and their families.

REFERENCES Airenne P (1981):X-linked hypohidrotic ectodermal dysplasia in Finland. R o c Finn Dent SOC77 (Suppl 1):1-106. Bixler D, Saksena S, Ward R (1988): Characterization of the face in hypohidrotic ectodermal dysplasia by cephalometric and anthropometric analysis. In Salinas CF, Opitz JM, Paul SW (eds):“Recent Advances in Ectodermal Dysplasia.” New York: Alan R. Liss, Inc., pp 197 -203. BMDP Software Inc. (1988): “BMDP Statistical Softwarc+User’s Digest, 4th Edition.” Los Angeles, CA: BMDP Statistical Software, Inc. Brodie AG, Sarnat BG (1942):Ectodermal dysplasia (anhidrotic type) with complete anodontia. Am J Dis Child 64:1046-1054. Bronowski J, Long WM (1952): Statistics of discrimination in anthropology. Am J Phys Anthropol 10:385-394. Brown GW (1984): Discriminant analysis. Am J Dis Child 138:395-400. Brownstein M (1973): Anhidrotic ectodermal dysplasia. J Baltimore Coll Dcnt Surg 28:65-72. Clarke A, Philippe DIM, Brown R, Harper PS (1987):Clinical aspects of X-linked hypohidrotic ectodermal dysplasia. Arch Dis Child 62:989 -996. Elejalde BR, Elejalde MM 11983): Pigmentary characteristics of the ectodermal dysplasias. J Clin Dysmorphol 15-8. Freire-Maia N, Pinheiro M (1984):“Ectodermal Dysplasia: A Clinical and Genetic Study.” New York: Alan R. Liss, Inc. Frias JL, Smith DW (1968): Diminished sweat pores in hypohidrotic ectodermal dysplasia. J Pediatr 72606-610. GorlinRJ, PindborgJJ, CohenMM(1976):“SyndromesoftheHeadand Neck.” 2nd Edition. New York: McGraw-Hill. Goodman RM, Gorlin R J (1983): “The Malformed Infant and Child.” New York: Oxford University Press. Hand DJ 11981): “Discrimination and Classification.” New York: John Wiley & Sons. Happle R, Frosch PJ (1985): Manifestation of the lines of Blaschko in women heterozygous for X-linked hypohidrotic ectodermal dysplasia. Clin Genet 27:468-471. Harbour J P (1981): “A Cephalometric Investigation of Hypohidrotic Ectodermal Dysplasia.” Indianapolis, IN: IU School of Dentistry. Howells WW (19731: “Cranial Variation in Man.” Cambridge, MA: Peabody Museum, Harvard University. Kerr CB; Wells RS, Cooper KE (1966): Gene effect in carriers of anhidrotic ectodermal dysplasia. J Med Genet 3:169-176. Kleinebrecht J , Degenhardt KH, Grubisic A, Gunther E, Svejcar J (1981): Sweat pore count in ectodermal dysplasias. Hum Genet 57:437-439. Lipshutz H (1963): Anhidrotic Ectodermal Dysplasia. J Albert Einstein Med Center 11:33-37. MacDermot KD, Winter RM, Malcolm S (1986): Gene localisation of X-linked hypohidrotic ectodermal dysplasia (C-S-T syndrome). Hum Genet 74:172-173. Nakata M, Koshiba H, Kazuhiro E, Nance W (1980):A geneticsstudy of anodontia in X-linked hypohidrotic ectodermal dysplasia. Am J Hum Genet 32:908-919. Oxnard C (1973): “Form and Pattern in Human Evolution.” Chicago, IL: University of Chicago Press.

112

Saksena and Bixler

Pinheiro M, Freire-Maia N (1979): Christ-Siemens-Touraine syndrome--A clinical and genetic analysis ofa large Braziliankindred: 111. Carrier detection. Am J Med Genet 4:129-134. %ed WB, Lopez LDA, Landing B (1970): Clinical spectrum of anhidrotic ectodermal dysplasia. Arch Dermatol 102:134-143. i D, yu ~ p (1987): l ~“A clinical ~ Atlas of Saksena s, Walker G, ~ lateralis,” N~~ york: ~l~~ R, Roentgenocepha1ometry in Liss, Inc. Saksena S (1989): A Clinical Atlas of RoentgenographicMeasurements in norna frontalis.” New York: Alan R. Liss, Inc., in press. Sarnat BG, Brodie AG, Kubacki WH (1953): Fourteen-Year %port of Facial Growth in Case of Complete Anodontia with Ectodermal Dysplasia. Am J Dis Child 86:162-169.

SAS Institute Inc (1986):“SAS User’s Guide: Statistics.” Cary NC: SAS Institute Inc. Settineri WMF, Salzano E’M, Freitas MJ (1976): X-linked anhidrotic ectodermal dysplasia with some unusual features. J Med Genet 13:212-216. Singh A, Jolly SS, Kaur S 11962):Hereditary ectodermal dysplasia. Br J Derm 74:34-37.

Soderholm AL, Kaitila I (1985):Expression of X-linked hypohidrotic ectodermal dysplasia in six males and in their mothers. Clin Genet 28:136-~144. Sofaer JA (1981a):A dental approach to carrier screening in X-linked hypohidrotic ectodermal dysplasia. J Med Genet 18:459-460. Sofaer JA (1981b): Hypodontia and sweat pore counts in detecting carriers of X-linked hypohidrotic ectodermal dysplasia. Br Dent J 151:327-330. SPSSX Inc (1988): “SPSS-X User’s Guide.” Chicago, IL: SPSS Inc. Thannhauser SJ (1936): Hereditary ectodermal dysplasia of the “anhydrotic type.” JAMA 106:908-910. Ward RE, Bixler D (1987): Anthropometric analysis of the face in hypohidrotic ectodermal dysplasia. Am J Phys Anthropol 7:453-458. Weech AA (1929): Hereditary ectodermal dysplasia (congenital ectadermal defect). Am J Dis Child 37:766-790. Zonana J , Clarke A, Sarfarazi M, Thomas NST, Roberts K, Marymee K, Harper PS (1988): X-linked hypohidrotic ectodermal dysplasia: Localization within the region Xqll-21.1 by linkage analysis and implication for carrier detection and prenatal diagnosis. Am J Hum Genet 43:75-85.

APPENDIX A. Variable Mean, Standard Deviation, Mean Difference, and F value of the Lateral (L) and Frontal (Pi Measurements of the Gene Carriers and Normal Relatives

Carrier Measurements

1 A Facial height Head height P-1 AP-CRO Facial height L-2 N-ME Midface height L-3 N-AN L-4 N-ZS L-5 N-ZI L-6 RiAN-PN L-7 OR/AN-PN Malar height L-8 a. OR-ZI b. ZS-ZI Upper alveolar height P-9 CNS-SD Lower alveolar height P-10 ID-ME Lower facial height L-11 AN-ME Posterior facial height L-12 S-GO 1B: Facial height & symmetry Apex height P-13 APiZY-ZY‘ P-14 APIMS-MS’ Orbital height P-15 a. ROLIZY-ZY’ b. RQRIZY-ZY‘ P-16 a. RQLIMS-MS’ b. RORIMS-MS‘ Maxillary height P-17 a. MXLIZY-ZY‘ b. MXWZY-ZY‘ Nasal shelf height P-18 a. NSLIZY-ZY’ b. NSWZY-ZY’ Infradental height P-19 IDIZY-ZY’ Menton height P-20 MEiZY-ZY‘ P-21 MEIMS-MS’

SD

Mean

SD

Mean difference (mm or degree)

93.2

8.0

100.1

6.8

-6.9

0.05

119.9

6.5

130.7

8.9

- 10.8

0.004

53.9 33.1 56.5 34.3 22.6

2.2 3.3 2.5 3.1 2.0

56.6 32.2 60.2 37.3 26.4

3.9 3.3 4.1 2.8 2.0

- 2.7

21.1 24.7

1.8 2.3

25.7 29.1

1.5 3.3

- 4.6

-4.4

0.0000 0.002

17.8

3.1

18.4

5.5

- 0.6

ns

32.1

3.9

34.3

4.8

- 2.2

ns

66.2

5.4

74.1

7.1

- 7.9

0.006

79.8

4.9

90.3

4.7

- 10.5

134.2 160.9

5.5 5.4

139.9 168.4

4.7 4.0

- 7.5

40.2 40.1 70.4 69.6

4.9 4.5 8.9 8.3

37.3 37.4 64.2 63.6

3.5 3.9 4.8 5.6

28.8 28.6

2.5 3.8

32.8 32.8

3.8 4.2

24.4

3.4 3.2

27.5 27.4

4.1 4.1

- 3.1

24.0

- 3.4

0.08 0.06

59.9

3.1

69.5

5.9

-9.6

0.0003

86.5 60.2

6.1 8.9

97.6 70.6

9.2 12.7

- 11.1

0.005 0.05

Mean

Normal

+ 0.9 - 3.7 - 3.0

- 3.8

- 5.7

+ 2.9 + 2.7 + 6.2 t 6.0 ~

4.0

- 4.2

~

10.4

Significance of F value

0.05 ns* 0.01 0.02 0.0001

0.0000

0.02 0.003

ns ns 0.05 0.05 0.009 0.05

(continued)

Facial Morphometry

113

APPENDIX A. Variable Mean, Standard Deviation, Mean Difference, and F value o f the Lateral (L) and Frontal (P) Measurements of the Gene Carriers and Normal Relatives (continued) Carrier Measurements 2A: Facial width Cranial width P-22 EU-EU’ Forehead width P-23 FT-FT’ P-24 RQ-RO’ Riorbital width P-25 LO-LO’ Interorbital width P-26 MO-MO’ Face width P-27 ZY-ZY’ Nasal width P-28 a. NC-NC’ b. NS-NS’ Maxillary width P-29 MX-MX’ Mandibular width P-30 GO-GO’ 2B: Facial depth & cranial base Midface deDth L-31 ERpt.-OR L-32 S-AN L-33 S-PN L-34 S-PG L-35 S-ME L-36 ERpt.-PG L-37 ERpt.-ME Palate depth L-38 PN-AN L-39 PT-AN Ramus depth L-40 AR-PG L-41 AR-GO Mandibular depth L-42 GO-PG L-43 GO-ME Mandibular angle L-44 AR-GO-ME Cranial length L-45 GL-OP Cranial base length L-46 S-N 12-47 S-BA L-48 N-AR L-49 N-BA Cranial base width P-50 MS-MS’ Cranial base angle 1,-51 N-S-BA 2C: Facial height-width-depth & symmetry P-52 a. AP-MOL b. AP-MOR P-53 a. MO-NCL b. M0’-NCR P-54 a. KC-NSL b. NC’-NSR P-55 a. NS-MXL b. NS‘-MXR P-56 a. MX-ZYL b. MX’-ZYR P-57 a. MX-GOL b. MX’-GOR P-58 a. ME-GOL b. ME-GOR

Mean

Normal

SD

Mean

SD

Mean difference (mm or degree)

Significance of F value

154.1

5.1

156.6

2.7

---2.5

ns

101.8 63.2

2.7 3.1

103.0 68.0

3.5 3.7

1,2 -4.8

ns 0.005

99.5

3.3

101.1

5.0

- 1.6

ns

24.0

2.5

25.2

2.1

1.2

ns

136.3

4.7

141.9

3.7

5.6

0.008

33.9 18.9

1.9

34.3

3.5

18.1

2.6 2.2

0.4 -L0.8

ns ns

57.8

2.8

63.7

3.1

-5.9

0.0003

102.6

4.1

107.8

6.9

- 5.2

0.02

76.1 85.7 50.7 125.7 126.9 121.4 120.9

3.5 3.3 2.6 4.0 4.2 6.4 6.3

81.7 92.9 55.3 138.3 139.5 134.7 134.2

4.6 2.8 2.9 4.7 5.4 6.3 6.5

5.6 7.2 -4.6 -- 12.6 - 12.6 - 13.3 13.3

50.4 54.7

2.9 2.9

54.7 60.5

2.1 2.7

108.9 46.2

6.2 5.6

120.3 52.8

4.7 4.7

78.9 75.7

5.6 4.7

84.7 81.0

5.4 5.1

124.6

6.3

125.7

4.5

196.2

2.4

203.4

7.0

74.0 45.0 97.3 107.7

3.0 4.6 4.4 3.6

77.1 49.5 104.4 113.5

2.1 6.6 4.9 6.5

114.8

5.0

121.6

5.6

129.8

7.4

126.8

4.2

113.7 113.0 38.6 38.5 10.6 10.0 20.9 20.2 48.3 48.8 41.2 40.1 57.5 56.1

8.2 8.1 3.3 3.3 1.3 1.0 1.4 1.8 2.9 3.2 3.8 4.5 3.4 3.2

120.2 119.3 40.4 40.3 11.1 10.3 23.9 23.3 51.1 51.2 43.4 44.5 60.2 61.5

7.0 7.6 3.4 2.5 1.5 1.0 2.3 2.1 4.5 4.8 4.6 4.7 6.3 6.4

~

~~

-~

~

~

--

~

4.3 5.8

0.0000 0.0005

11.4

0.0000 0.005

~

~

~

- 6.6

~

-

5.8 5.3

- 1.1

7.2

~

- 3.1

~

-

~

0.01

6.5 1.8

- 1.8

0.5

~

- 0.3

3.0 3.1 --2.8 - 2.4 2.2 - 4.4 - 2.5 - 5.4 ~

.-

0.003

6.8

- 6.3 ~

ns

7.1 5.8

t 3.0”

~

0.02 0.01

0.009 0.06 0.001 0.01

- 4.5

~

0.003 0.0000 0.0004 0.@0@0 0.0000 0.0000 0.0000

11s

0.07 0.09 ns ns ns ns 0.002 0.002 ns ns ns 0.05 ns 0.02

(continued)

114

Saksena and Bixler

APPENDIX A. Variable Mean, Standard Deviation, Mean Difference, and F value of the Lateral (L) and Frontal (P) Measurements of the Gene Carriers and Normal Relatives (continued) Carrier Normal Mean difference (mm Significance SD Mean SD or degree) of F value Measurements Mean

3:

P-59 a. ZY-ROL b. ZY'-ROR P-60 a. ZY-NCL b. ZY'-NCR P-61 a. ZY-NSL b. ZY'-NSR C. SD-ZYL d. SD-ZYR P-62 a. ZY-GOL b. ZY'-GOR P-63 a. ME-ZYL b. ME-ZYK P-64 a. ZY-MSL b. ZY'-MSR Facial profile Forehead angle L-65 Mftpt.-GL-N Nasal angle L-66 GL-N-R Malar-nasal angles L-67 ZS-N-R L-68 ZI-N-R Malar inclination at cranial base L-69 S-NIZS-ZI L-70 S-N/OR-ZI Maxillary and mandibular inclinations at cranial base L-71 S-N-Apt L-72 S-N-Bpt Facial angles L-73 Apt.-N-Bpt. L-74 N-Apt.-PG

~

+ 1.6 + 1.0

54.3 54.5 55.1 54.9 65.3 64.7 82.6 81.7 66.8 64.9 113.4 109.8 33.2 32.7

3.7 4.7 3.1 3.4 3.8 3.4 5.3 4.7 7.1 6.6 8.8 8.8 5.9 5.4

52.7 53.5 58.1 58.4 68.9 68.1 85.6 86.3 73.2 73.4 122.4 121.3 30.5 28.8

2.6 3.5 2.5 2.4 3.2 3.2 5.5 6.7 6.2 5.4 9.1 9.2 4.4 5.1

152.7

4.3

144.2

9.4

129.1

10.1

122.5

6.8

52.9 60.2

7.2 7.5

58.9 65.6

6.5 5.1

-

42.9 52.8

6.8 6.9

47.1 56.8

5.4 6.3

-4" -4"

79.3 76.9

4.1 3.6

83.4 81.1

3.0 3.7

-

2.31 1.45

2.0 6.4

2.34 1.84

2.4 7.5

- 0.03" -3.29"

3.0 - 3.5 - 3.6 - 3.4 -3.0 - 4.6 6.4 - 8.5 9.0 - 11.5 +2.7 + 3.9 -

~

~

ns

ns 0.03 0.02 0.03 0.03 ns 0.09 0.04 0.006 0.04 0.01 ns 0.05

+ 8.5" + 7"

0.09

6" 5"

0.01 0.02

~

~

4" 4"

0.5

ns ns 0.008 0.01 ns ns

"F value not significant.

APPENDIX B. Morphometric Method: The 3 Steps Involved in the Morphornetric Identification of Individuals Step 1. Take the following 3 linear measurements (to the nearest 0.5 mm) on a PA radiograph: a. XI (IDIZY-ZY') b. X2 (MX-MX') C. X3 (ME-GOR)

Step 2. Make adjustments for age and sex effects for each measurement (Y1, Y2 & Y3) in Step 1 by using the following HED-specific regression equations:" Y, (adjusted) = XI + 0.5580 (25-age) - 0.0030 (625-12) Y2 (adjusted) = Xz + 0.5734 (25-Age) - 0.0042 (625-12) Ys (adjusted) = X3 + 0.7504 (25-Age) - 0.0052 (625-12) Step 3. Enter the three adjusted variable values (Y1, Yz, and Ys into the following discriminant function (Eq. 2): Discriminant score = 0.1438Y1 + 0.3429Y2 + o.2052Y3 - 42.202 If the value of the discriminant score for a given individual is below zero, the individual is classified as a gene carrier; if above zero, the individual is normal (see Table 11; Fig. 3). "Age is actual age (months in decimal point); if age is equal to or greater than 25 years, then age = 25, if actual age is less than 25 years, then the males = 1.1 and females = 1.2; I2 = Age2 x sex.