HLA Raymond
R.
Antigens
and
Schizophrenia
Crowe, MD; John S. Thompson, MD; Ronald Flink; Bernice Weinberger, RN
\s=b\ We typed 45 schizophrenic patients for 35 HLA antigens and compared their frequencies with 1,263 population controls. No significant differences between schizophrenics and controls were found. When the schizophrenics were subtyped, a significant (P < .05) excess of Aw26 was found among the hebephrenics, compared with the population controls. When the published literature on schizophrenia-HLA associations was surveyed, none of the reported associations were found to be consistent across studies. Some possible explanations for the heterogeneity among studies are discussed and it is concluded that an association between schizophrenia and any of the HLA antigens has not yet been demonstrated. (Arch Gen Psychiatry 36:231-233, 1979)
demonstration of associations between human and a number of diseases1 has a search for similar associations with psychiatric disease. The literature now contains four such studies on schizophrenia that suggest a number of associations. Cazzullo et al2 and Smeraldi and his coworkers3 4 have published several reports indicating a negative association between schizophrenia and histocom¬ patibility antigen AIO. Others have reported positive asso¬ ciations with A93 and with A28.6 Bennahum et al7 found a relative decrease in A1J, B7, and B8 and a relative increase in Bwl7 in their schizophrenic sample. The present inves¬ tigation was undertaken in an attempt to replicate these findings in an Iowa population that is relatively homoge¬ neous with respect to HLA haplotype frequencies.
Thehistocompatibility antigens (HLA) encouraged
SUBJECTS AND METHODS The present series consists of 45 schizophrenic patients from the Iowa Psychiatric and Veterans Administration hospitals meeting
Accepted for publication Feb 1, 1978. From the Departments of Psychiatry (Dr Crowe and Ms Weinberger) and Medicine (Dr Thompson) and the Tissue Typing Laboratory (Mr Flink), University of Iowa College of Medicine, Iowa City. Reprint requests to Department of Psychiatry, University of Iowa College of Medicine, 500 Newton Rd, Iowa City, IA 52242 (Dr Crowe).
Downloaded From: by a East Carolina University User on 01/15/2019
a diagnosis of schizophrenia.8 The 45 schizo¬ phrenics were subtyped according to published criteria8 as 20 hebephrenics and 14 paranoids, with the subtype of the remaining 11 undetermined. Diagnoses were made without knowledge of the patients' HLA haplotypes. Their ages ranged from 20 to 50 years, with a mean of 28.4 years. Sixteen of the subjects were women and
research criteria for
29,
men.
Histocompatibility antigen typing was performed using a modified Amos lymphocytotoxicity method.1" Seventeen antigens were tested at the A locus and 18 at the locus. The antigens tested are enumerated in Table 1. Histocompatibility antigen frequencies in the schizophrenic sample were compared with an Iowa population sample of 1,262 subjects. Differences were tested with a standard 2 test and probabilities obtained were multiplied by 35 (the number of antigens tested) to correct for type 1 errors.
RESULTS In Table 1, the diagnostic groups are compared with the Iowa control sample. Four antigens, Aw9(24), Aw30, Aw32, and B14, were found more frequently among the schizo¬ phrenics at the .05 probability level in the uncorrected data, but when the data were corrected for the number of antigens tested, no differences approached statistical difference. When the schizophrenic sample was subdivided into hebephrenic and paranoid, a single positive finding emerged: Awl0(26) was found in 15% of the hebephrenics, compared with 2.6% of the Iowa population, a finding significant at the .05 level after correction. No other findings approached statistical significance. Because of the stringent criterion of multiplying the significance level by the number of antigens tested, a true association would have to be very strong in order to survive the significance testing. Table 2 compares the present study with the four already published with respect to deviations that were found to be significant at the .05 level on uncorrected significance testing. If a true association exists, it might be detectable in this manner, with congruous findings among the studies suggesting that the
Table 1.—HLA
Schizophrenia Hebephrenla (N = 45) ( = 20)
HLA A series 1
9 9
Table 2.—HLA
Frequencies
Antigen Frequencies in Schizophrenics (S) and Controls (C)
Control
Paranoid
(N = 14)
(
=
1,263) Present
31 49 20 0
30 40 20 0
21 50 36 0
30.5 57.6 26.6 11.9
(w23)_2_0_0_1.6 18 20f _21_8j>
9(w24) 10
2
7
0
4.9
10(w25)_0 0_0_1.0 10ÍW26)_7_15_0_2J3 _
11 w19 28 29 w30 w31 w32 w33 Blank series 5
11 0 7 11 9 2 9 0 22 11
29 18 33 2 13
12 13 14 w15 16
13 2
w17
4
18 w21 w22 27 w35 w37 w40 4C w39 Blank
2 7
•Percents
9 13 9 2 11
0 0
20
0 10 15 5
0 0
9.7 4.2
7
6.3 6.1 3.2 2.3 3.3 0.3 18.9
14
0 0 7 0 36
10 0 20
14
5 35 10 30
11.9 34.0 20.9
36 21 43 0
5
10 10 5 5
28.4 3.2
21 21 0
6.0 12.8 3.7 8.0 2.9 2.3 4.4
7
5
0
10 10 20
0
10 5 15 0
7
14.3
0 0 0
0.6 13.0 0.7
0 10
0
0.1
14
24.4
to 200 because two alíeles less than .05.
sum
tProbability
15 0
7 7
are
typed
7.9
at each locus.
association is not chance. It is clear from Table 2 that no such trends have developed. The studies of Bennahum et al7 have suggested significant deviations with All, B7, B8, and Bwl7 that have not appeared in the other studies. Eberhard et al5 reported significant deviations with A5 (now designated B5), A9, and Awl9 that again failed to find support among the other studies that examined for these antigens. However, A9 has subsequently been shown to be heterogeneous, containing at least two other specific¬ ities designated w23 and w24. In the present study, w24 showed a positive association at the .05 level on the uncorrected data, perhaps lending some support to the findings of Eberhard et al. Ivanyi et al6 found a highly significant positive associa¬ tion with A28, but this has not been confirmed by other studies. The present study found positive associations with Aw9(24), Aw30, Aw32, and B14 in the uncorrected data.
Downloaded From: by a East Carolina University User on 01/15/2019
Study 45
Ivanyl
22 9
A11 Aw19 A28 Aw30 Aw32 B5 B7 B8 B14 Bw15 Bw17 Bw27
10
Eberhard étal5
étal'
1,262 144 386 148
20 A9t A9(w24) 18 9 A10t
Bw35
Smeraldi et al"
1,200 47
31
28
24
23
13 13
28
23
16
11 9
Bennahum et al'
1,263
34
15
30
16
38
102
13
19
29 18 13 13
12 34 21
30
24
13 13
9 11
10 22 14
23 17
21 34 17
11
16
11
19
13
12
28 26 2 .
8 3
24 20
18
7
.
13
14
10
22
16
16
•Rounded to the nearest percent. Associations significant at the .05 level on uncorrected data are indicated in boldface type. fA9 and A10 were thought to be single antigenic specificities when most of these studies were reported. They have now been subdivided into A9, Aw9(23), Aw9(24) and A10, Aw10(25), and Aw10(26) respectively. There¬ fore, in comparing studies, the subspecificities have been pooled under A9 and A10.
Aside from the positive association with A9 noted above, data on only the last two of these antigens are available from other studies and in neither case is the association supported. Finally, Smeraldi et al3-4 reported a positive association with All and negative associations with Bwl5, Bw35, and AIO. When their data were corrected for multiple tests, the AIO association fell just short of statis¬ tical significance in their first series, but was replicated on a second and independent series, and in Table 2 both series are pooled, with an additional 22 patients they subsequent¬ ly added. This is the only finding thus far to be replicated on an independent sample and thus deserves comment. The AIO antigen is found in 8.5% of the Iowa population and thus would have been expected in only four of our 45 schizophrenics, rendering the present study indeterminate for this association. Likewise, the expectation among the 47 schizophrenics studied by Eberhard et al would have been four, making that study indeterminate, as well. The sample of Ivanyi et al was drawn from a population with a 16% frequency of AIO, making the hypothesis testable with his 148 schizophrenics. However, AIO was found more frequently among schizophrenics than among controls. COMMENT
Perhaps the most striking aspect of the HLA studies is variability in findings from one investigation to anoth¬ er. Although sampling variation may explain some of this variability, other explanations have been proposed that could account for much of it.11 If schizophrenia is genetical¬ ly heterogeneous, a true association with one subgroup the
might exist that would be obscured by the other subgroups present in the sample. The proportion of subtypes may
one series to the next, leading to lack of replication. To this end, it is important that patient populations be subtyped in an attempt to minimize this source of heterogeneity. Subtyping the present series led to a significant (P < .05) association between hebephrenia and Awl0(26). A second source of inconsistency is allelic heterogeneity. Gene frequencies of the various HLA alíeles vary greatly
vary from
between ethnic groups and an alíele may be so rare in a given population as to make detection of a true association unlikely. Moreover, the disease may be associated with one alíele in one ethnic group and a different alíele at the same locus in another. Third, serologie heterogeneity may contribute to the confusion. Antiserums once thought to be specific for a single antigen have been found to react with two or more antigens on subsequent testing. This problem could have contributed to the discrepancy between the present data and that of Eberhard et al as noted previously. Given the magnitude of the problems involved, it is not surprising to find discrepancies between the studies. It would appear that the best method for resolving these difficulties would be to study large kindreds with several members affected. This strategy has been employed in the study of ragweed pollen allergy, a disease characterized by locus alíeles, but with the alíele a strong association with involved varying from one family to another.12 This approach, which assures disease as well as allelic homo¬ geneity within families, could help to clarify the discrepan¬ cies raised by the population studies. Nevertheless, some generalizations may be made concerning the current data. Bach and VanRood11 note that HLA-disease associations can be characterized by magni-
tude of the 2, degree of homogeneity of the studies, and size of the relative risk (ie, the increase in risk of the development of the disease given a particular HLA alíele). Thus characterized, the associations fall into two groups: (1) those with overwhelmingly large 2, homogeneous studies, and large relative risks >4, and (2) a second group in which the 2 are small, studies are heterogeneous, and relative risks are two or less. It is clear that schizophrenia falls into the second category. Generalizations from the HLA-disease literature raise other problems for schizophrenia, as well. Most associations have been with the locus, whereas the strongest reported association with schizophrenia is with an A locus alíele. Moreover, the established associations are with susceptibil¬ ity to the disease (positive association), rather than resistance (negative association), as has been found with AIO. Thus, despite a number of positive studies, no consistent associations between schizophrenia and any HLA alíele have been uncovered. The strongest candidate is a negative association with an A locus alíele, both of these points running counter to the trends that have emerged from other HLA-disease associations. Positive associations with A9 and with Aw9(24) (formerly included under A9) have now been observed in two studies and need to be studied further. The method employed thus far, that of population studies, introduces a number of sources of heterogeneity that could account for the wide discrepancies between studies. The best approach for resolving these differences would be to look for associations within large families, but to date no such studies have been reported with schizophre¬ nia to our knowledge. This study was supported in part Research Fund.
by a grant from the Iowa Mental Health
References 1. Sasazuki T, McDevitt HO, Grumet FC: The association between genes in the major histocompatibility complex and disease susceptibility. Ann Rev Med 28:425-452, 1977. 2. Cazzullo CL, Smeraldi E, Penati G: The leucocyte antigenic system HLA as a possible genetic marker of schizophrenia. Br J Psychiatry 125:25-27, 1974. 3. Smeraldi E, Bellodi L, Cazzullo CL: Further studies on the major histocompatibility complex as a genetic marker for schizophrenia. Biol
Psychiatry 11:655-661, 1976. 4. Smeraldi E, Bellodi L, Scorza-Smeraldi R, et al: HLA-SD antigens and schizophrenia: Statistical and genetical considerations. Tissue Antigens 8:191-196, 1976.
5. Eberhard G, Franzen G, L\l=o"\wB: Schizophrenia susceptibility and HLA antigen. Neuropsychobiology 1:211-217, 1975. 6. Ivanyi D, Zemek P, Ivanyi P: HLA antigens in schizophrenia. Tissue
Downloaded From: by a East Carolina University User on 01/15/2019
Antigens 8:217-220,
1976. 7. Bennahum DA, Troup GM, Rada RT, et al: HLA antigens in schizophrenia and manic depressive mental disorders. Clin Res 25:114A, 1977. 8. Feighner JP, Robins E, Guze SB, et al: Diagnostic criteria for use in psychiatric research. Arch Gen Psychiatry 26:57-63, 1972. 9. Tsuang MT, Winokur G: Criteria for subtyping schizophrenia: Clinical differentiation of hebephrenic and paranoid schizophrenia. Arch Gen Psychiatry 31:43-47, 1974. 10. Ray JG, Hare DB, Kayhoe DE: Manual of Tissue Typing Techniques, publication 76-545. Bethesda, Md, US Department of Health, Education, and Welfare, 1976. 11. Bach FH, VanRood JJ: The major histocompatibility complex: Genetics and biology. N Engl J Med 295:927-936, 1976. 12. Levine BB, Stember RH: Ragweed hay fever: Genetic control and linkage to HL-A haplotypes. Science 178:1201-1203, 1972.
R.
Antigens
and
Schizophrenia
Crowe, MD; John S. Thompson, MD; Ronald Flink; Bernice Weinberger, RN
\s=b\ We typed 45 schizophrenic patients for 35 HLA antigens and compared their frequencies with 1,263 population controls. No significant differences between schizophrenics and controls were found. When the schizophrenics were subtyped, a significant (P < .05) excess of Aw26 was found among the hebephrenics, compared with the population controls. When the published literature on schizophrenia-HLA associations was surveyed, none of the reported associations were found to be consistent across studies. Some possible explanations for the heterogeneity among studies are discussed and it is concluded that an association between schizophrenia and any of the HLA antigens has not yet been demonstrated. (Arch Gen Psychiatry 36:231-233, 1979)
demonstration of associations between human and a number of diseases1 has a search for similar associations with psychiatric disease. The literature now contains four such studies on schizophrenia that suggest a number of associations. Cazzullo et al2 and Smeraldi and his coworkers3 4 have published several reports indicating a negative association between schizophrenia and histocom¬ patibility antigen AIO. Others have reported positive asso¬ ciations with A93 and with A28.6 Bennahum et al7 found a relative decrease in A1J, B7, and B8 and a relative increase in Bwl7 in their schizophrenic sample. The present inves¬ tigation was undertaken in an attempt to replicate these findings in an Iowa population that is relatively homoge¬ neous with respect to HLA haplotype frequencies.
Thehistocompatibility antigens (HLA) encouraged
SUBJECTS AND METHODS The present series consists of 45 schizophrenic patients from the Iowa Psychiatric and Veterans Administration hospitals meeting
Accepted for publication Feb 1, 1978. From the Departments of Psychiatry (Dr Crowe and Ms Weinberger) and Medicine (Dr Thompson) and the Tissue Typing Laboratory (Mr Flink), University of Iowa College of Medicine, Iowa City. Reprint requests to Department of Psychiatry, University of Iowa College of Medicine, 500 Newton Rd, Iowa City, IA 52242 (Dr Crowe).
Downloaded From: by a East Carolina University User on 01/15/2019
a diagnosis of schizophrenia.8 The 45 schizo¬ phrenics were subtyped according to published criteria8 as 20 hebephrenics and 14 paranoids, with the subtype of the remaining 11 undetermined. Diagnoses were made without knowledge of the patients' HLA haplotypes. Their ages ranged from 20 to 50 years, with a mean of 28.4 years. Sixteen of the subjects were women and
research criteria for
29,
men.
Histocompatibility antigen typing was performed using a modified Amos lymphocytotoxicity method.1" Seventeen antigens were tested at the A locus and 18 at the locus. The antigens tested are enumerated in Table 1. Histocompatibility antigen frequencies in the schizophrenic sample were compared with an Iowa population sample of 1,262 subjects. Differences were tested with a standard 2 test and probabilities obtained were multiplied by 35 (the number of antigens tested) to correct for type 1 errors.
RESULTS In Table 1, the diagnostic groups are compared with the Iowa control sample. Four antigens, Aw9(24), Aw30, Aw32, and B14, were found more frequently among the schizo¬ phrenics at the .05 probability level in the uncorrected data, but when the data were corrected for the number of antigens tested, no differences approached statistical difference. When the schizophrenic sample was subdivided into hebephrenic and paranoid, a single positive finding emerged: Awl0(26) was found in 15% of the hebephrenics, compared with 2.6% of the Iowa population, a finding significant at the .05 level after correction. No other findings approached statistical significance. Because of the stringent criterion of multiplying the significance level by the number of antigens tested, a true association would have to be very strong in order to survive the significance testing. Table 2 compares the present study with the four already published with respect to deviations that were found to be significant at the .05 level on uncorrected significance testing. If a true association exists, it might be detectable in this manner, with congruous findings among the studies suggesting that the
Table 1.—HLA
Schizophrenia Hebephrenla (N = 45) ( = 20)
HLA A series 1
9 9
Table 2.—HLA
Frequencies
Antigen Frequencies in Schizophrenics (S) and Controls (C)
Control
Paranoid
(N = 14)
(
=
1,263) Present
31 49 20 0
30 40 20 0
21 50 36 0
30.5 57.6 26.6 11.9
(w23)_2_0_0_1.6 18 20f _21_8j>
9(w24) 10
2
7
0
4.9
10(w25)_0 0_0_1.0 10ÍW26)_7_15_0_2J3 _
11 w19 28 29 w30 w31 w32 w33 Blank series 5
11 0 7 11 9 2 9 0 22 11
29 18 33 2 13
12 13 14 w15 16
13 2
w17
4
18 w21 w22 27 w35 w37 w40 4C w39 Blank
2 7
•Percents
9 13 9 2 11
0 0
20
0 10 15 5
0 0
9.7 4.2
7
6.3 6.1 3.2 2.3 3.3 0.3 18.9
14
0 0 7 0 36
10 0 20
14
5 35 10 30
11.9 34.0 20.9
36 21 43 0
5
10 10 5 5
28.4 3.2
21 21 0
6.0 12.8 3.7 8.0 2.9 2.3 4.4
7
5
0
10 10 20
0
10 5 15 0
7
14.3
0 0 0
0.6 13.0 0.7
0 10
0
0.1
14
24.4
to 200 because two alíeles less than .05.
sum
tProbability
15 0
7 7
are
typed
7.9
at each locus.
association is not chance. It is clear from Table 2 that no such trends have developed. The studies of Bennahum et al7 have suggested significant deviations with All, B7, B8, and Bwl7 that have not appeared in the other studies. Eberhard et al5 reported significant deviations with A5 (now designated B5), A9, and Awl9 that again failed to find support among the other studies that examined for these antigens. However, A9 has subsequently been shown to be heterogeneous, containing at least two other specific¬ ities designated w23 and w24. In the present study, w24 showed a positive association at the .05 level on the uncorrected data, perhaps lending some support to the findings of Eberhard et al. Ivanyi et al6 found a highly significant positive associa¬ tion with A28, but this has not been confirmed by other studies. The present study found positive associations with Aw9(24), Aw30, Aw32, and B14 in the uncorrected data.
Downloaded From: by a East Carolina University User on 01/15/2019
Study 45
Ivanyl
22 9
A11 Aw19 A28 Aw30 Aw32 B5 B7 B8 B14 Bw15 Bw17 Bw27
10
Eberhard étal5
étal'
1,262 144 386 148
20 A9t A9(w24) 18 9 A10t
Bw35
Smeraldi et al"
1,200 47
31
28
24
23
13 13
28
23
16
11 9
Bennahum et al'
1,263
34
15
30
16
38
102
13
19
29 18 13 13
12 34 21
30
24
13 13
9 11
10 22 14
23 17
21 34 17
11
16
11
19
13
12
28 26 2 .
8 3
24 20
18
7
.
13
14
10
22
16
16
•Rounded to the nearest percent. Associations significant at the .05 level on uncorrected data are indicated in boldface type. fA9 and A10 were thought to be single antigenic specificities when most of these studies were reported. They have now been subdivided into A9, Aw9(23), Aw9(24) and A10, Aw10(25), and Aw10(26) respectively. There¬ fore, in comparing studies, the subspecificities have been pooled under A9 and A10.
Aside from the positive association with A9 noted above, data on only the last two of these antigens are available from other studies and in neither case is the association supported. Finally, Smeraldi et al3-4 reported a positive association with All and negative associations with Bwl5, Bw35, and AIO. When their data were corrected for multiple tests, the AIO association fell just short of statis¬ tical significance in their first series, but was replicated on a second and independent series, and in Table 2 both series are pooled, with an additional 22 patients they subsequent¬ ly added. This is the only finding thus far to be replicated on an independent sample and thus deserves comment. The AIO antigen is found in 8.5% of the Iowa population and thus would have been expected in only four of our 45 schizophrenics, rendering the present study indeterminate for this association. Likewise, the expectation among the 47 schizophrenics studied by Eberhard et al would have been four, making that study indeterminate, as well. The sample of Ivanyi et al was drawn from a population with a 16% frequency of AIO, making the hypothesis testable with his 148 schizophrenics. However, AIO was found more frequently among schizophrenics than among controls. COMMENT
Perhaps the most striking aspect of the HLA studies is variability in findings from one investigation to anoth¬ er. Although sampling variation may explain some of this variability, other explanations have been proposed that could account for much of it.11 If schizophrenia is genetical¬ ly heterogeneous, a true association with one subgroup the
might exist that would be obscured by the other subgroups present in the sample. The proportion of subtypes may
one series to the next, leading to lack of replication. To this end, it is important that patient populations be subtyped in an attempt to minimize this source of heterogeneity. Subtyping the present series led to a significant (P < .05) association between hebephrenia and Awl0(26). A second source of inconsistency is allelic heterogeneity. Gene frequencies of the various HLA alíeles vary greatly
vary from
between ethnic groups and an alíele may be so rare in a given population as to make detection of a true association unlikely. Moreover, the disease may be associated with one alíele in one ethnic group and a different alíele at the same locus in another. Third, serologie heterogeneity may contribute to the confusion. Antiserums once thought to be specific for a single antigen have been found to react with two or more antigens on subsequent testing. This problem could have contributed to the discrepancy between the present data and that of Eberhard et al as noted previously. Given the magnitude of the problems involved, it is not surprising to find discrepancies between the studies. It would appear that the best method for resolving these difficulties would be to study large kindreds with several members affected. This strategy has been employed in the study of ragweed pollen allergy, a disease characterized by locus alíeles, but with the alíele a strong association with involved varying from one family to another.12 This approach, which assures disease as well as allelic homo¬ geneity within families, could help to clarify the discrepan¬ cies raised by the population studies. Nevertheless, some generalizations may be made concerning the current data. Bach and VanRood11 note that HLA-disease associations can be characterized by magni-
tude of the 2, degree of homogeneity of the studies, and size of the relative risk (ie, the increase in risk of the development of the disease given a particular HLA alíele). Thus characterized, the associations fall into two groups: (1) those with overwhelmingly large 2, homogeneous studies, and large relative risks >4, and (2) a second group in which the 2 are small, studies are heterogeneous, and relative risks are two or less. It is clear that schizophrenia falls into the second category. Generalizations from the HLA-disease literature raise other problems for schizophrenia, as well. Most associations have been with the locus, whereas the strongest reported association with schizophrenia is with an A locus alíele. Moreover, the established associations are with susceptibil¬ ity to the disease (positive association), rather than resistance (negative association), as has been found with AIO. Thus, despite a number of positive studies, no consistent associations between schizophrenia and any HLA alíele have been uncovered. The strongest candidate is a negative association with an A locus alíele, both of these points running counter to the trends that have emerged from other HLA-disease associations. Positive associations with A9 and with Aw9(24) (formerly included under A9) have now been observed in two studies and need to be studied further. The method employed thus far, that of population studies, introduces a number of sources of heterogeneity that could account for the wide discrepancies between studies. The best approach for resolving these differences would be to look for associations within large families, but to date no such studies have been reported with schizophre¬ nia to our knowledge. This study was supported in part Research Fund.
by a grant from the Iowa Mental Health
References 1. Sasazuki T, McDevitt HO, Grumet FC: The association between genes in the major histocompatibility complex and disease susceptibility. Ann Rev Med 28:425-452, 1977. 2. Cazzullo CL, Smeraldi E, Penati G: The leucocyte antigenic system HLA as a possible genetic marker of schizophrenia. Br J Psychiatry 125:25-27, 1974. 3. Smeraldi E, Bellodi L, Cazzullo CL: Further studies on the major histocompatibility complex as a genetic marker for schizophrenia. Biol
Psychiatry 11:655-661, 1976. 4. Smeraldi E, Bellodi L, Scorza-Smeraldi R, et al: HLA-SD antigens and schizophrenia: Statistical and genetical considerations. Tissue Antigens 8:191-196, 1976.
5. Eberhard G, Franzen G, L\l=o"\wB: Schizophrenia susceptibility and HLA antigen. Neuropsychobiology 1:211-217, 1975. 6. Ivanyi D, Zemek P, Ivanyi P: HLA antigens in schizophrenia. Tissue
Downloaded From: by a East Carolina University User on 01/15/2019
Antigens 8:217-220,
1976. 7. Bennahum DA, Troup GM, Rada RT, et al: HLA antigens in schizophrenia and manic depressive mental disorders. Clin Res 25:114A, 1977. 8. Feighner JP, Robins E, Guze SB, et al: Diagnostic criteria for use in psychiatric research. Arch Gen Psychiatry 26:57-63, 1972. 9. Tsuang MT, Winokur G: Criteria for subtyping schizophrenia: Clinical differentiation of hebephrenic and paranoid schizophrenia. Arch Gen Psychiatry 31:43-47, 1974. 10. Ray JG, Hare DB, Kayhoe DE: Manual of Tissue Typing Techniques, publication 76-545. Bethesda, Md, US Department of Health, Education, and Welfare, 1976. 11. Bach FH, VanRood JJ: The major histocompatibility complex: Genetics and biology. N Engl J Med 295:927-936, 1976. 12. Levine BB, Stember RH: Ragweed hay fever: Genetic control and linkage to HL-A haplotypes. Science 178:1201-1203, 1972.
