S.Mj
`
`Linkage of Early-Onset Familial Breast Cancer
`to Chromosome 17q21
`
`JEFF M. HALL, MING K. LEE, BETH NEWMAN, JAN E. MoRRow,
`LEE A. ANDERSON, BING HUEY, MARY-CLAIuRE KING
`
` on August 18, 2009
`
`www.sciencemag.org
`
`Downloaded from
`
`unavoidable epidemiologic realities. The disease is common, but
`only a small proportion of cases in the general population are
`attributable to inherited susceptibility. Thus, families may have
`multiple cases of breast cancer without inherited susceptibility, and
`"sporadic" cases may occur even in families with inherited disease. In
`addition, the disease is not completely penetrant among susceptible
`persons, with expression depending on gender, age, and nongenetic
`risk factors. Finally, both epidemiological and molecular evidence
`suggests heterogeneity. We have tested simultaneously for genetic
`linkage and heterogeneity of breast cancer in families, and our
`results suggest both the presence of a gene for early-onset breast
`cancer on chromosome 17q21 and linkage heterogeneity of the
`disease.
`Families and inhertance of susceptibility. Our genetic analysis
`is based on 23 extended families with 146 cases of breast cancer
`(Figs. 1 and 2). All persons in our analysis are Caucasian and from
`a variety of original ancestries. The 329 participating relatives now
`live in, and were therefore sampled from, 40 states of the United
`States, Puerto Rico, Canada, the United Kingdom, and Colombia.
`These families share the epidemiological features that are character-
`istic of familial, versus sporadic, breast cancer (2): younger age at
`diagnosis, frequent bilateral disease, and more frequent occurrence
`of disease among men.
`Our statistical model for the inheritance ofsusceptibility to breast
`cancer was derived from our previous complex segregation analysis
`of a population-based series of 1500 families with breast cancer (4).
`Inherited susceptibility to breast cancer in that series could be fully
`explained by a rare autosomal dominant allele with a major effect on
`risk: risk of breast cancer in genetically susceptible women was
`estimated to be 0.37 by age 40, 0.66 by age 55, and 0.82 over the
`entire lifetime. In contrast, risk of breast cancer in women without
`genetic susceptibility was estimated to be 0.004 by age 40, 0.028 by
`age 55, and 0.081 over the entire lifetime. Females less than 15 years
`of age and all males had a negligible risk (less than 0.001). The
`estimated proportion of breast cancer cases in the sample that were
`attributable to inherited susceptibility was only 4 percent, the great
`majority of cases resulting purely from somatic events. Among
`younger patients, however, the proportion ofinherited cases is likely
`to be considerably higher. Disease allele frequencies (q) between
`0.004 and 0.02 yield virtually identical results; those for q equals
`0.01 are described.
`Definition of the breast cancer phenotype. For any complex
`disease, it is essential to adequately define the phenotype, the
`inheritance of which will be traced in families. Real linkages can be
`missed and spurious linkages suggested either by defining the
`phenotype too broadiy (so that persons without inherited suscepti-
`bility to disease are mistakenly categorized as affected) or simply by
`making errors in diagnosis. To minimiz errors in diagnosis, we
`
`SCIENCE, VOL. 250
`
`Human breast cancer is usually caused by genetic alter-
`ations of somatic cells of the breast, but occasionally,
`susceptibility to the disease is inherited. Mapping the
`genes responsible for inherited breast cancer may also
`allow the identification ofearly lesions that are critical for
`the development of breast cancer in the general popula-
`tion. Chromosome 17q21 appears to be the locale of a
`gene for inherited susceptibility to breast cancer in fami-
`lies with early-onset disease. Genetic analysis yields a lod
`score (logarithm of the likelihood ratio for linkage) of
`5.98 ror linkage of breast cancer susceptibility to DI 7S74
`in early-onset families and negative lod scores in families
`with late-onset disease. Likelihood ratios in favor of
`linkage heterogeneity among families ranged between
`2000:1 and greater than 106:1 on the basis ofmultipoint
`analysis of four loci in the region.
`
`HU UMAN DISEASE GENES CAN BE LOCATED BY LINKAGE
`analysis of families in which the incidence of the disease is
`high. Linkage analysis can reveal the chromosomal loca-
`tion of the genes of interest by identifying polymorphic genetic
`markers of known location that are coinherited with the disease in
`families (1). Among the common cancers, breast cancer is particu-
`larly suited for this approach, because family history ofthe disease is
`a significant risk factor in all populations; epidemiological evidence
`consistently indicates that a woman's risk of breast cancer is in-
`creased by the occurrence ofthe disease in her mother or sisters. The
`younger the ages at diagnosis ofher relatives, the greater the increase
`in a woman's risk (2).
`The transformation of breast ductal epithelial cells to malignant
`growth results from alterations in their DNA that may be either
`inherited or somatic (3). Mapping genes for familial breast cancer is
`important because alterations at the same loci may also be respon-
`sible for sporadic disease. Individuals with inherited susceptibility to
`breast cancer are completely asymptomatic for decades before the
`onset of disease; the effects of critical inherited alterations are thus
`latent for an extended period. Among women with no inherited
`susceptibility to the disease, these same alterations may be the initial
`lesions of breast tumorigenesis, with disease expression being simi-
`larly dependent on subsequent genetic alterations or tumor-promot-
`ing steps.
`Mapping genes for human breast cancer has been complicated by
`
`The authors are at the School of Public Health, University of California, Berkeley, CA
`94720.
`
`1684
`
`GeneDX 1028, pg. 1
`
`

`

` on August 18, 2009
`
`www.sciencemag.org
`
`Downloaded from
`
`reviewed existing pathology records ofall family members on whom
`breast surgery had been performed. For deceased persons reported
`by their relatives to have had breast cancer, but for whom no
`pathology records were available, we obtained hospital records or
`death certificates. For living subjects who had not undergone breast
`surgery, we relied on self-report of no breast cancer; for deceased
`persons with no history of breast surgery, we relied on death
`* certificates and reports of relatives. The affected phenotype was
`defined as all histologic types of invasive breast cancer. No other
`cancer sites were included (5).
`Statistical methods. Four approaches to evaluating linkage were
`applied:
`1) Lod scores (logarithms of the likelihood ratios for linkage) for
`linkage of individual markers to disease were estimated with the
`LIPED program; multipoint mapping was performed with the
`LINKAGE program; homogeneity of recombination fractions was
`evaluated with the B test for two-point linkage data and direct
`comparison of likelihoods for multipoint data (1, 6). Homogeneity
`of linkage was tested for all families; the sample was not a priori
`subdivided. Linkage analyses were based on an autosomal dominant
`model with the age- and sex-specific risks described above for
`hypothetically susceptible and nonsusceptible individuals. (LIPED
`was modified to incorporate four liability classes for each genotype.)
`The risk group (liability dass) for each woman was defined by her
`age at diagnosis ofbreast cancer, at death ifdeceased without breast
`cancer, or at most recent interview if living and unaffected. All men
`were assigned to the lowest risk group.
`2) In determining a plausible model for inheritance of suscepti-
`bility to a complex disease, it is always possible that the underlying
`genetic model may be correct, but that penetrance may not be
`
`accurately estimated. Therefore, we also tested for linkage by
`including disease information only for the affected relatives (that is,
`the breast cancer cases) in each family; all unaffected subjects were
`assigned to the lowest risk class, so that only their marker informa-
`tion was incorporated. Autosomal dominance was still assumed, and
`lifetime risk of sporadic disease was not altered.
`3) In order to evaluate linkage without imposing any specific
`genetic model, we counted alleles shared by descent for affected pairs
`of female relatives. This analysis was possible because one marker
`was highly polymorphic. Pairs ofaffected relatives were sisters (first-
`degree relatives), aunt-niece and grandmother-granddaughter (sec-
`ond-degree relatives), and first cousins (third-degree relatives).
`Relative pairs were stratified by the average age of breast cancer
`diagnosis for the pair.
`4) We applied the affected-pedigree-member method to evaluate
`sharing of alleles by state among the breast cancer cases in each
`family (7). Only cases whose marker genotypes could be determined
`with certainty were included in the analysis of affected pedigree
`members.
`Typing of DNA polymorphisms. For each of the 329 informa-
`tive relatives, we obtained 35 milliliters of fresh blood and prepared
`immortalized lymphocytes by Epstein-Barr viral transformation (8).
`Genomic DNA was prepared as described (9). Probes were labeled
`by random primer extension (10) and hybridized to DNA according
`to standard procedures (11). Parentage was confirmed by consistent
`inheritance of 183 polymorphic markers. Markers at chromosome
`17q21 include D17S74, a VNTR (variable number of tandem
`repeats) defined by the probe CMM86 and HinfI; D17S40, defined
`by the probe LEW 101 and Msp I; D17S41, defined by the probe
`LEW 102 and Pst I; and D17S78, defined by the probe p131 and
`
`2
`
`dx44
`
`3
`
`D-S
`
`Ov
`
`65
`AE
`
`BD
`
`dx45
`BA
`
`dx39
`BD
`
`Ov
`
`72
`DA
`
`EA
`
`04
`
`9
`DD
`
`C
`
`FG
`
`dx3l
`
`dx45 dx36
`ED
`EH
`
`0 G o *l 4
`
`71
`DD
`
`EH
`
`BC
`
`dx2B
`(B
`
`AH
`
`dx29
`EB
`
`48
`HC
`
`43
`HC
`
`dx37 40
`El
`HF
`
`dx4l
`BA
`
`dx23
`BE
`
`39
`DE
`
`52
`A E
`
`dx42 dx39
`AA
`AA
`
`BC
`
`dx3l
`AC
`
`dx25
`EA
`
`*0b 4-C
`
`72
`
`BC
`
`4
`
`dx45 dx4l
`
`4 &
`
`dx 45
`CB
`
`9 2
`CE
`
`4
`
`BC
`
`866
`CE
`
`77
`CB
`
`32
`(B-)
`
`DE
`
`DE
`
`-4
`
`dx 45
`CB
`
`Fig. 1. Breast cancer families 1 to
`females with
`Solid circles,
`7.
`breast cancer; open cirdes, fe-
`males without breast cancer; open
`males without breast
`squares,
`cancer. The age given for each
`woman is age at (first) breast can-
`cer diagnosis (dx) if affected, age
`at death if deceased (deceased in-
`dividuals are represented by diag-
`onal lines through symbols), or
`age at most recent interview if
`alive without breast cancer. Al-
`leles of D17S74 are shown for all
`families and are lettered sequen-
`tially within each family from
`largest to smallest fragment size.
`Alleles in parentheses are based
`on reconstructed genotypes.
`
`21 DECEMBER 1990
`
`dx47
`BE
`
`dx39
`BE
`
`CD
`
`60
`AC
`
`62 dx41
`EC
`BD
`
`51
`ED
`
`dx43
`B
`
`dx40
`BE
`
`CE
`
`iz
`
`dx32 31
`EE
`IC
`
`6
`
`dx 60
`
`AE
`
`AD
`
`82
`Af
`
`dx43
`CE
`
`dx 46
`CA
`
`dx45
`CA
`
`dx46
`(CE)
`
`88
`
`dx4l
`CB
`
`44
`EB
`
`dx5l
`{D-)
`
`dx4D
`(C-)
`
`DG
`
`dx42
`(C-)
`
`73
`CC
`
`68
`CC
`
`CB
`
`DD
`
`59
`EC
`
`dx59
`CD
`
`dx 40
`
`CF
`
`dx3D
`BD
`
`dx36
`CD
`
`36
`BD
`
`7
`
`0a
`
`77
`CD
`
`dx44
`OD
`
`AC
`
`dx5B
`OD
`
`{C- I
`
`6 6 4b 6 6
`
`dx46
`BC
`
`dx48
`BC
`
`42
`BC
`
`dx38
`BD
`
`6- j
`
`54
`DA
`
`dx33
`BC
`
`RESEARCH ARTICLES
`
`1685
`
`GeneDX 1028, pg. 2
`
`

`

` on August 18, 2009
`
`Msp I (12). On the basis ofanalysis ofthe CEPH families, the order
`and approximate recombination distances of the markers are:
`17cen-D17S78-(0.10)-Dl7S41-(0.06)-D17S74-(0.12)-D17S40
`-17qter (13).
`D17S74, a highly polymorphic VNTR with heterozygosity great-
`er than 0.90, is extremely useful for linkage analysis but presents
`technical challenges that are common to VNTRs. Linkage analysis is
`
`critically sensitive both to errors in assigning genotypes and to
`marker allele frequencies. The kcngths of the Hinfl restriction
`fragments that define D17S74 alleles are continuously distributed
`between 1 and 5 kilobase pairs (kb). For our analyses, D17S74
`alleles were identified by analyzing DNA samples from all members
`of a family on the same Southern blot, placing relatives with
`fragments ofsimilar size adjacent to one another (14). "Population"
`
`a
`
`9
`
`dx42
`
`10
`
`a-*
`
`dx65
`AB
`
`CD
`
`61
`75
`x ZB
`
`dx37
`(G-)
`
`(G-)
`
`dx48
`(A-)
`
`dx56 dx53 dx 32
`ZB
`
`,B
`
`I|
`
`dxZ8
`(D-) AB
`
`dx62
`
`dx5O
`(AE)
`
`(DE)
`
`dx66
`
`94
`Df
`
`CC
`
`dx65
`(D-) BE
`
`dx45
`DB
`
`CH
`
`82 dx49 72
`[A
`DE
`DE
`
`71 dx61 dx56
`DA
`EE
`
`70 dxS0 51
`DC
`FC
`DC
`
`dx35
`DR
`
`11
`
`38
`DlH
`
`33 dx28
`BC
`DC
`
`12
`
`13
`
`6a
`
`73
`AF
`
`(C-)
`
`F -
`
`Ef
`
`dx53
`BD
`
`dx45
`
`dx49
`
`dx65
`AD
`
`dx 40
`AD
`
`57
`AD
`
`-638
`
`BC
`
`dx64
`CD
`
`DE
`
`6 d4
`
`73
`AB
`
`dx56
`CB
`
`46
`
`dx57
`AD
`
`-E
`
`DE
`
`56
`BC
`
`6
`
`dx4O
`AD
`
`46 6
`
`dx44
`AD
`
`44
`AC
`
`i b (b) bb
`
`dx32
`CE
`
`35
`BE
`
`33
`BE
`
`31
`CD
`
`30
`CE
`
`14
`
`( B
`
`dx>,70
`(AC)
`
`dx63
`CD
`
`66 dxSSdx33
`GG
`GG
`
`dxS9
`AC
`
`dx57dx52 54
`FD
`ED
`EB
`
`dx43
`CE
`
`dx 48
`(D-)
`
`dx49
`CD
`
`GH
`
`dx 71
`CB
`
`75
`AD
`
`dx6l
`CB
`
`dx62 dx 42
`DD
`CD
`
`dx44
`CD
`
`CB
`
`www.sciencemag.org
`
`Downloaded from
`
`15
`
`16
`
`dx3O
`DlH
`
`27
`DG
`
`21
`ClH
`
`17
`
`dx 44
`BD
`
`45
`BD
`
`38
`CD
`
`is
`
`dx77 8
`De
`CD
`
`( DF) AE
`
`d65 72
`
`79 71
`
`(CD)
`
`eS
`
`dx66
`(AO)
`
`lb _ 4) i
`
`dx66 dx35
`
`dx66
`DF
`
`71
`ED
`
`dx63
`AC
`
`62
`AD
`
`52
`ED
`
`*4 1
`
`dx44
`AD
`
`dx41
`AC
`
`66
`BD
`
`62
`RC
`
`dx58
`DC
`
`dx4O dx32
`FE
`DE
`
`dx65
`HJ
`
`77
`GK
`
`EJ
`
`dx32dx35 37
`EG
`EG
`JK
`
`19
`
`dx47 51 dx38 38
`xAB
`BR
`B18
`
`AB
`
`dx47 34
`1 : x
`
`21
`
`(AC)
`
`A
`
`dxS0 56
`AK
`FK
`
`dx48
`GK
`
`FK
`
`dx 61
`IE
`
`20
`
`dxS1
`
`dx87
`
`91
`
`dx>70
`
`79
`
`dx47
`
`88
`
`dx45
`
`6-6-
`it
`4,~~Hb
`
`dx52
`AC
`
`AF
`
`dx58
`EF
`
`BE
`
`BE
`
`dx7l
`BE
`
`70
`BD
`
`DF
`
`BD
`
`63
`BC
`
`59
`BD
`
`dxSB
`RD
`
`41 dx36
`BC
`DR
`
`22
`
`23
`
`dx7O
`
`40
`AC
`
`dx72
`AC
`
`dxS4
`AB
`
`dx57
`AB
`
`dx57
`BC
`families. D17S78 genotypes for
`Genotys for D17S74 are shown for al
`family 8 and D17S40 genotypes for family 19 are also shown.
`
`0T,9dx 79
`6 6 '6 4
`
`70
`AA
`
`66
`AC
`
`dx49
`BC
`
`dx48
`BC
`
`dx57 dx62
`75 dx58dx62
`EG
`EG
`EE
`DD
`Cf
`Fig. 2. Breast cancer families 8 to 23. Notation as for Fig. 1, with solid
`squares in families 16 and 19 representing males with breast cancer.
`
`1686
`
`SCIENCE, VOL. 250
`
`GeneDX 1028, pg. 3
`
`

`

` on August 18, 2009
`
`www.sciencemag.org
`
`Downloaded from
`
`frequencies of the D17S74 alleles in this sample were estimated by
`selecting subjects from different families whose D17S74 fragments
`appeared to be ofsimilar size on the basis oftheir "family blots," and
`then analyzing the DNA from these unrelated persons in neighbor-
`ing lanes on the same blots. Some samples were included several
`times in order to identify distortions in the gels. These "population
`blots" were analyzed without reference to sample numbers, in order
`to determine which alleles could be consistently distinguished.
`D17S74 had more than 30 distinguishable fragment lengths-and
`hence more than 30 different alleles-in our sample, nine of which
`occurred more than once among unrelated individuals, at frequen-
`cies ranging from 0.07 to 0.13. The other D17S74 alleles were only
`represented once in our sample, but because extremely rare marker
`allele frequencies can have a major influence on estimates of lod
`scores and the T statistic (7), apparently unique alleles were each
`assigned the frequency 0.03.
`Results of linkage and heterogeneity analysis in the breast
`cancer families. For the 23 families as a group, homogeneity of
`linkage of breast cancer to D17S74 could be rejected at P equals
`0.01. Multipoint analysis oflinkage in the interval D17S78-D17S41-
`D17S74-D17S40 yielded likelihood ratios in favor of heterogeneity
`oflinkage among the 23 families between 2000:1 and 1.4 x 106 to
`1. After adjusting for heterogeneity among all families, the maxi-
`mum two-point lod score is + 3.28 at recombination distance of
`0.014 from D17S74, with disease linked to this locus in 40 percent
`of the families (Figs. 1 and 2).
`Heterogeneity of linkage in these families appears explicable by
`age of disease onset. Breast cancer is linked to markers in this
`chromosomal region specifically in families with early-onset breast
`cancer. Among the seven families with a mean age of breast cancer
`diagnosis less than or equal to 45, the two-point lod score for
`linkage of D17S74 and breast cancer is +5.98 at a distance of 0.001
`recombination units, with a 95 percent confidence interval of 0.001
`to 0.09 (Table 1). In contrast, total lod scores for the families with
`late-onset disease are negative. It is characteristic of linkage in the
`presence of heterogeneity that a modest lod score (in this case
`+2.35) for all families, ignoring heterogeneity, at a fairly large
`recombination distance (0.20 recombination units) masks two
`curves, one with a more positive lod score in the linked families
`(+5.98) at a smaller recombination fraction (0.001 to 0.09) and the
`other negative (15).
`Two-point lod scores for all four markers in this chromosomal
`region suggest that a gene for susceptibility in the early-onset
`families is likely to be within approximately 10 percent recombi-
`nation of D17S74 (Table 2). Multipoint analysis of the four-
`marker interval yields a maximum lod score of + 5.41 near D17S74
`for the earliest-onset families (Table 2). Again, total lod scores for
`families with older ages at diagnosis are negative throughout the
`interval.
`Linkage of breast cancer to D17S74 was also evaluated on the
`basis of only the individuals with breast cancer in each family. For
`this analysis, all women without breast cancer and all men were
`assigned to the lowest risk group. For the 23 families as a group, the
`P value for homogeneity of linkage is 0.06. For the families with
`average age at diagnosis less than or equal to 45, the maximum lod
`score is +4.69 at close linkage, with a 95 percent confidence interval
`for the recombination fraction of 0.001 to 0.10. Lod scores at close
`linkage to D17S74 are -2.19 for families 8 to 15 and -5.22 for
`families 16 to 23.
`Analysis of alleles shared by descent among related pairs of
`women with breast cancer also suggested linkage of early-onset
`breast cancer to D17S74 (Table 3). In families 1 to 7, all three classes
`of relatives shared more alleles by descent than expected by chance.
`Even in families 8 to 23, there was evidence for increased identity by
`
`Table 1. Lod scores for linkage of breast cancer to D17S74, chromosome
`17q21. For each family, M is the mean age of diagnosis of breast cancer.
`
`Family
`
`M
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`
`32.7
`37.2
`37.3
`39.8
`42.6
`44.2
`45.4
`47.0
`47.4
`47.6
`49.3
`50.2
`50.4
`51.4
`51.8
`52.0
`53.5
`53.6
`55.8
`56.4
`58.7
`59.4
`63.3
`
`Recombination fraction
`0.10
`0.20
`0.30
`
`+1.89
`+0.35
`+0.29
`+0.91
`-0.25
`+1.06
`+0.58
`+0.02
`+0.03
`-0.06
`-0.41
`-0.03
`-0.09
`-0.18
`-0.08
`-0.56
`+0.04
`-0.38
`-0.93
`-1.01
`+0.50
`-0.13
`-0.02
`
`+1.38
`+0.21
`+0.19
`+0.64
`-0.08
`+0.73
`+0.40
`+0.02
`+0.06
`-0.08
`-0.13
`-0.02
`-0.02
`+0.01
`-0.02
`-0.20
`+0.07
`-0.18
`-0.45
`-0.56
`+0.34
`+0.04
`+0.00
`
`+0.82
`+0.09
`+0.09
`+0.35
`+0.00
`+0.41
`+0.21
`+0.01
`+0.04
`-0.08
`-0.03
`-0.01
`-0.03
`+0.06
`-0.01
`-0.07
`+0.05
`-0.07
`-0.20
`-0.28
`+0.18
`+0.05
`+0.00
`
`0.40
`
`+0.28
`+0.02
`+0.03
`+0.11
`+0.03
`+0.14
`+0.05
`+0.00
`+0.01
`-0.05
`-0.00
`-0.00
`-0.04
`+0.04
`-0.00
`-0.02
`+0.01
`-0.02
`-0.05
`-0.11
`+0.05
`+0.02
`+0.00
`
`ZZ at
`0.001
`+2.36
`+2.86
`+3.26
`+4.40
`+3.90
`+5.28
`+5.98
`+5.98
`+5.67
`+5.63
`+4.12
`+4.06
`+3.65
`+3.00
`+2.65
`-0.06
`-0.19
`-0.94
`-3.50
`-5.21
`-4.56
`-5.41
`-5.48
`
`0.001
`
`+2.36
`+0.50
`+0.40
`+1.14
`-0.50
`+1.38
`+0.70
`+0.00
`-0.31
`-0.04
`-1.51
`-0.06
`-0.41
`-0.65
`-0.35
`-2.71
`-0.13
`-0.75
`-2.56
`-1.71
`+0.65
`-0.85
`-0.07
`
`descent among sisters with early-onset disease, but little or no
`evidence for increased identity by descent for second- and third-
`degree relatives. Finally, the affected-pedigree-member analysis of
`alleles shared by state among all individuals with breast cancer in
`each family also suggested linkage of early-onset breast cancer to
`markers in this chromosomal region (16).
`These analyses defined families with early-onset disease as those
`with an average age at breast cancer diagnosis of less than or equal
`to 45. However, "early-onset' could be defined in a variety ofways.
`The cumulative lod scores for linkage ofbreast cancer to D17S74, as
`families with increasing age at disease onset are included in the
`analysis, are indicated in Fig. 3 and the right-hand column ofTable
`1. Cumulative lod scores are above 5.0 for families with an average
`age at diagnosis ofless than 48, remain positive for families with an
`average age at diagnosis of less than 52, and then drop sharply. As
`specific alternative ways of defining early-onset breast cancer, we
`defined families with early-onset disease to be those in which (i)
`most breast cancers in each family were diagnosed by age 50
`(families 1 to 7, 9, 10, 12, 13, 21), or (ii) most breast cancers in each
`family were diagnosed by age 45 (families 1 to 6, 10), or (iii) the
`average age at breast cancer diagnosis was less than 48, the mean for
`all the cases in the sample (families 1 to 10). The critical results for
`all definitions were the same: maximum lod scores for linkage of
`breast cancer to this chromosomal region are between 5.2 and 5.7 in
`the early-onset families and negative in the late-onset families, Pnd
`the disease gene appears to be within about 0.10 recombination
`units of D17S74.
`Other risk factors for breast cancer in the families. To deter-
`mine whether the linked gene is expressed in the presence of any
`specific background of nongenetic risk factors for breast cancer, we
`compared breast cancer risk factors for women in families with
`apparent linkage ofbreast cancer to chromosome 17q versus women
`in families with evidence against this linkage. Ages at first pregnan-
`cy, number ofchildren, prevalence offertility problems, exposure to
`x-rays, use of oral contraceptives, and ages at menopause were
`similarly distributed in "linked" and "unlinked" families, after
`adjusting for age and birth cohort of the women. Similarly, preva-
`
`21 DECEMBER 1990
`
`RESEARCH ARTICLES 1687
`
`GeneDX 1028, pg. 4
`
`

`

`Table 3. Identity by descent of two, one, or zero alleles of D17S74
`among pairs of related individuals with breast cancer. Mean onset is the
`average age of breast cancer diagnosis for the related pair. Family numbers
`refer to Figs. 1 and 2. Second-degree (20) relatives are aunt-niece or
`grandmother-granddaughter pairs; third-degree (30)
`are first
`relatives
`cousins.
`
`Mean onset of
`related pair
`(years)
`
`Sisters
`
`20
`
`30
`
`Two
`
`One
`
`Zero
`
`One
`
`Zero
`
`One
`
`Zero
`
`s45
`46-55
`.56
`
`s45
`46-55
`
`'45
`46-55
`-56
`
`14
`5
`3
`
`9
`0
`
`5
`5
`3
`
`All families
`0
`0
`2
`Families 1 to 7
`0
`0
`Families 8 to 23
`0
`0
`2
`
`4
`8
`6
`
`2
`0
`
`2
`8
`6
`
`19
`9
`7
`
`17
`2
`
`2
`7
`7
`
`5
`2
`7
`
`0
`0
`
`5
`2
`7
`
`8
`4
`3
`
`6
`1
`
`2
`3
`3
`
`4
`4
`10
`
`2
`0
`
`2
`4
`10
`
` on August 18, 2009
`
`www.sciencemag.org
`
`Downloaded from
`
`higher prevalence of sporadic cases in families with late-onset
`inherited disease. Loci responsible for disease in families with
`late-onset disease may be identifiable by continued simultaneous
`analysis of linkage and heterogeneity.
`Candidate genes on chromosome 17q. The ultimate goal of
`gene mapping of human traits is to move from a known chromo-
`somal location to identification of the crucial gene and characteriza-
`tion of its critical alterations. This region of chromosome 17q
`includes several plausible candidate genes (21). A gene for a
`truncated form of the human epidermal growth factor receptor
`[her2; MIM 164870 (Mendelian Inheritance in Man)] is identical to
`erbb2 or neu (MIM 190150). The gene her2 acts as an oncogene in
`NIH 3T3 cells and is amplified in many primary breast tumors; her2
`amplification is associated with poor prognosis at least for node-
`positive tumors (22). Other candidate genes in this region include
`that for estradiol-17P dehydrogenase (edhbl7; MIM 264300),
`which is the enzyme that catalyzes the conversion of estrone to
`estradiol; the homeobox 2 gene (hox2; MIM 142960), which is
`critical for murine embryological development; nm23, whose expres-
`sion is associated with lymph node metastasis in primary breast
`carcinomas; the gene for retinoic acid receptor a (rara; MIM
`180240), a protein that binds the possible anticarcinogen retinoic
`acid; and wnt3 (MIM 165330), one ofthe integration sites activated
`by the mouse mammary tumor virus and which is homologous to
`
`Fig. 3. Linkage ofDI 7S74
`to breast cancer in families,
`based on the autosomal
`dominant model described
`in the test. Cumulative lod
`scores (IZ) are shown for
`all families for which the
`mean age of breast cancer
`diagnosis (M) is less than
`or equal to the age repre-
`sented on the x-axis. Total
`lod scores are greater than
`or equal to 5.0 for families
`with M ' 48 and greater
`than or equal to 3.0 for
`families with M ' 50. Each
`dot above the x-axis repre-
`sents a family with that par-
`ticular mean onset age.
`
`6-
`
`4
`
`2
`
`30
`
`40
`
`50
`
`60
`
`-2
`
`Age of onset (years)
`
`lences of specific cancers at other sites did not differ among women
`in "linked" and "unlinked" families, although male breast cancers
`occurred only in "unlinked" families. The only observed difference
`between women in "linked" versus "unlinked" families was age at
`breast cancer diagnosis.
`Linkage analysis in families and loss of heterozygosity in
`tumors. Comparisons of breast tumor tissue with normal tissue
`from the same individual have suggested that chromosomes lp, 3p,
`llp, 13q, 16q, or 17p (or some combination) may harbor genes
`that are important for breast tumor progression (17). An earlier
`linkage analysis from our group, based on fewer and less-informative
`families, suggested (with a modest lod score) that a gene for familial
`premenopausal breast cancer or ovarian cancer is present on chro-
`mosome 16q (18). We have excluded the other regions suggested by
`loss of heterozygosity in tumors as locales of genes for inherited
`susceptibility to breast cancer (9, 19).
`A gene or genes on chromosome 5q appear to be responsible for
`inherited forms of colon cancer and quite possibly for the initial
`somatic lesions of other colon cancers, with genes on 12p (K-ras),
`17p (p53), 18q (DCC), and possibly elsewhere being responsible
`for subsequent invasion and metastasis (20). This pattern ofmultiple
`sequential events, determined by alterations on more than one
`chromosome, may also apply to breast cancer.
`Negative lod scores in the families with late-onset breast cancers
`may reflect any or all of the following: the existence of a different
`locus or loci responsible for inherited susceptibility in these families;
`the chance occurrence of some families with multiple cases; or a
`
`Table 2. Linkage analysis of breast cancer to four markers on chromosome 17q by mean age (M) of breast cancer diagnosis.
`
`Marker
`
`Families 16 to 23 (M s 52)
`0.30
`0.20
`0.10
`
`0.40
`
`0.001
`
`Families 1to 7 (M s 45)
`0.30
`
`0.20
`
`0.10
`
`Two-point lod scores at five recombination fractions
`Familes 8 to 15 (46 s M s 51)
`0.30
`0.20
`
`0.40
`
`0.001
`
`0.10
`
`0.40
`
`-0.65
`-1.12
`+4.83
`+ 1.01
`
`-0.16
`-0.71
`+3.47
`+0.63
`
`-0.04
`-0.36
`+ 1.97
`+0.30
`
`+0.00
`-0.14
`+0.65
`+0.07
`
`+0.36
`+0.10
`-3.33
`-0.49
`
`+0.95
`+0.51
`-0.80
`-0.12
`
`+0.81
`+0.43
`-0.18
`+0.01
`
`+0.48
`+0.26
`-0.05
`+0.06
`
`+0.14
`+0.08
`-0.04
`+0.05
`
`-4.18
`-2.73
`-8.13
`-2.42
`
`-1.25
`-1.03
`-2.49
`-0.71
`
`-0.44
`-0.59
`-0.94
`-0.25
`
`Lod scores based on multipoint analysis of the interval (on recombination units)
`S74
`S41
`0.160
`0.100
`
`0.184
`
`0.208
`
`0.232
`
`0.256
`
`0.040
`
`0.060
`
`0.080
`
`0.120
`
`0.140
`
`-0.04
`-0.16
`-0.12
`-0.00
`
`-0.12
`-0.35
`-0.34
`-0.05
`
`S40
`0.280
`
`0.001
`
`-0.84
`-1.54
`+5.98
`+ 1.36
`
`S78
`.000
`
`+2.83
`-0.30
`-6.70
`
`D17S78
`D17S41
`D17S74
`D17S40
`
`Families
`
`1 to 7
`8 to 15
`16 to 23
`
`1688
`
`0.020
`
`+3.09
`-0.07
`-5.80
`
`+3.30
`+0.01
`-5.51
`
`+3.47
`+0.03
`-5.52
`
`+3.57
`-0.05
`-5.89
`
`+3.41
`-0.20
`-6.98
`
`+4.46
`-1.58
`-6.60
`
`+5.24
`+4.60
`-9.14
`-2.71
`-7.94 -15.21
`
`+5.41
`-5.61
`-8.94
`
`+5.24
`-4.24
`-6.79
`
`+4.96
`-3.36
`-5.58
`
`+4.48
`-2.78
`-5.03
`
`SCIENCE) VOL. 250
`
`GeneDX 1028, pg. 5
`
`

`

` on August 18, 2009
`
`www.sciencemag.org
`
`Downloaded from
`
`the Drosophila winglessness locus (23). If alterations in any of these
`genes are responsible for inherited breast cancer, polymorphisms at
`the critical locus may be in linkage disequilibrium with the disease in
`the early-onset families.
`
`REFERENCES AND NOTES
`1. J. Ott, Analysis ofHuman Genetic Linkage (Johns Hopkins Univ. Press, Balimore,
`1985).
`2. R. Ottman, M. C. Pike, M.-C. King, J. T. Casagrande, B. E. Hendcrson, Am. J.
`Epidemiol. 123, 15 (1986); N. L. Petrakis, V. L. Ernster, M.-C. King, in Cancer
`Epidemiology and Prevention, D. Schottenfeld and J. F. Fraumeni, Jr., Eds.
`(Saunders, Philadelphia, 1982), pp. 855-870; A. G. Schwartz, M.-C. King, S. H.
`Belle, W. A. Satariano, G. M. Swanson, J. Natt. Cancer Inst. 75, 665 (1985).
`3. A. G. Knudsen, Cancer Res. 45, 1437 (1985).
`4. B. Newman, M. A. Austin, M. Lee, M.-C. King, Proc. Natl. Acad. Sci. U.S.A. 85,
`3044 (1988).
`5. Four women with endometrial cancer (from families 1, 4, and 9) and two women
`with thyroid canccr (from families 5 and 13) werc omitted entirely from the
`analysis. In family 3, two women have breast and ovarian cancer and arc defined as
`affected at the ages of their breast cancer diagnoses; two other womcn died with
`ovarian cancer only and could not be induded.
`6. J. Ott, Ann. Hum. Genet. 47, 311 (1983); G. M. Lathrop, J. M. Laloucl, C. Julier,
`J. Ott, Am. J. Hum. Genet. 37, 482 (1985); N. Risch, ibid. 42, 353 (1988).
`7. D. E. Weeks and K. Lange, Am. J. Hum. Genet. 42, 315 (1988).
`8. L. Louie and M.-C. King, ibid. 41, A174 (abstr.) (1987).
`9. J. M. Hall et al., ibid. 44, 577 (1989).
`10. A. P. Feinberg and B. Vogelstein, Anal. Biochem. 132, 6 (1983); ibid. 137, 266
`(1984).
`11. T. Maniatis, F. F. Fritsch, J. Sambrook, Molecular Cloning: A Laboratory Manual
`(Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, 1982).
`12. Y. Nakamura et al., Genomics 2, 302 (1988); J. Luty et al., Nucleic Acids Res. 16,
`6250 (1988).
`13. Centre d'Etude du Polymorphisme Humain, CEPH Version 3 Database: Lod Scores
`and Recombination Estimates (CEPH, Paris, 1989).
`14. An example of the difficulties in typing VNTR alleles is presented by family 18, in
`which allele K is alleged to appear in cousins without pedigre evidence for identity
`by descent in intervening relatives. This assignment, and others like it, were made
`after analyzing several samples from the relatives on the same Southern blot,
`genotyping them without reference to their sample numbers or positions in the
`
`pedigree, and detcrmining that the bands were indistinguishable among the four
`samples.
`15. L. L. Cavalli-Sforza and M.-C. King, Am. J. Hum. Genet. 38, 599 (1985).
`16. For the 23 families as a whole, the method of affected pedigree members yields a
`T value of 1.13 (P = 0.13) for the T statistic unadjusted for allele frequencies, T
`equals 4.26 (P < 0.001) for the inverse square root weighting function, and T
`equals 5.49 (P < 0.001) for the reciprocal weighting function. The T statistic
`among the younger families was consistently significant, with T values of 2.60,
`5.18, and 5.78, and empirical P values of 0.012, 0.001, and 0.003 for the three
`weighting functions, respectively.
`17. P. Devilee et al., Genomics 5, 554 (1989); M. Genuardi, H. Tsihira, D. E.
`Anderson, G. F. Saunders, Am. J. Hum. Genet. 45, 73 (1989); I. U. Ali, R.
`Lidereau, C. Theillet, R. Callahan, Science 238, 185 (1987); C. Lundberg, L.
`Skoog, W. K. Cavenee, Proc. Natl. Acad. Sci. U.S.A. 84, 2372 (1987); C. Coles
`et al., Lancet 336, 761 (1990); T. Sato et al., Am. J. Hum. Genet. 47, A16 (abstr.)
`(1990).
`18. M.-C. King, R. C. P. Go, R. C. Elston, H. T. Lynch, N. L. Petrakis, Science 208,
`406 (1980); M.-C. King et al., J. Natl. Cancer. Inst. 71, 463 (1983).
`19. A. M. Bowcock, J. M. Hall, J. M. Hebert, M.-C. King, Am. J. Hum. Genet. 46,
`12 (1990); J. M. Hall et al., Genomics 6, 181 (1990).
`20. B. Vogelstein et al., Science 244, 207 (1989); E. R. Fearon and B. Vogelstein, Cell
`61, 759 (1990).
`21. V. A. McKusick, Mendelian Inheritance in Man (Johns Hopkins Univ. Press,
`Baltimore, ed. 10, 1990); E. Solomon and D. F. Barker, Cytogenet. Cell Genet. 51,
`319 (1989).
`22. L. Coussens et al., Science 230, 1132 (1985); N. C. Popescu, C. R. King, M. H.
`Kraus, Genomics 4, 362 (1989); P. P. Di Fiore et al., Science 237, 178 (1987); M.
`J. van de Vijver and R. Nusse, Biochim. Biophys. Acta, in press; P. Devilee and C.
`J. Comelisse, Cancer Surv., in press.
`23. V. Luu-The et al., Mol. Endocrinol. 4, 268 (1990); G. Bevilacqua, M. E. Sobel, L.
`A. Liotta, P. S. Steeg, Cancer Res. 49, 5185 (1989); A. M. Rosengard et al., Nature
`342, 177 (1989); H. Roelink, E. Wagenaar, S. Lopes da Silva, R. Nusse, Proc.
`Natl. Acad. Sci. U.S.A. 87,4519 (1990); M. Petkovich, N. J. Brand, A. Krust, P.
`Chambon, Nature 330, 444 (1987); N. Brand et al., Nature 332, 850 (1988).
`24. We thank Y. Nakamura, M. Litt, and D. Barker for probes; the CEPH for
`informnation on genetic markers; N. L. Petrakis and S. Wolman for help with
`interpretation of breast tumor pathology; J. Ott and N. Risch for statistical advicc;
`E. Jones, S. Rowell, P. Zuppan, and J. Crandall for technical assistance; and E.
`Anderson and the Breast Cancer Task Force ofNCI for encouragement. Supported
`by NIH grant CA27632.
`
`12 September 1990; accepted 10 November 1990
`
`AAAS-Newcomb Cleveland Prize
`To Be Awarded for an Article or a Report Published in Science
`
`The AAAS-Newcomb Cleveland Prize is awarded to the
`