Patterns of inheritance in familial ALS Abstract—We investigated 185 families with ALS for evidence of anticipation and mitochondrial inheritance. Although initial analysis demonstrated significant anticipation of age at death between generations in patients with familial ALS, further analysis demonstrated features of regression to the mean, suggesting that the perceived differences are the result of bias. In addition, there was no evidence of an effect of preferential maternal inheritance, which would have supported transmission of mitochondrial DNA mutations. NEUROLOGY 2005;64:1628 –1631 Marcus Bradley, MRCP; Lloyd Bradley, MRCP; Jackie de Belleroche, PhD, FRCPath; and Richard W. Orrell, MD, FRCP Anticipation is the tendency for a disease to demonstrate an earlier age at onset or increasing severity of symptoms in successive generations.1,2 Apparent anticipation has been reported in five Japanese3,4 and three Italian5 families with ALS and copper/zinc superoxide dismutase (SOD1) mutations. An excess number of male-transmitting grandparents and great-grandparents, suggesting selection against a large expansion in the female gamete, has been reported.6 We found no significant change using a similar analysis.7 Bulbospinal muscular atrophy is a related motor neuron disease, with a trinucleotide repeat expansion in the androgen receptor gene, although anticipation is relatively rare.8 Mitochondrial abnormalities have been demonstrated in different tissues of patients with ALS.9 Mitochondrial DNA disorders are usually maternally inherited and a maternal effect on inheritance would be expected, if mutations of mitochondrial DNA are causative. Using a database of families with ALS,10 we investigated patterns of inheritance for evidence of anticipation and preferential maternal inheritance. Methods. We studied 185 families, with at least two individuals having typical features of ALS.10 Individual transmissions between generations were identified. Intergenerational values were compared between generation 1 (G1) and generation 2 (G2) for each individual transmission. G1 comprised fathers, mothers, uncles, and aunts. G2 comprised sons, daughters, nephews, and nieces. The differences in age at onset, duration of disease, and age at death between G1 and G2 were calculated. For first-degree transmissions, the number of cases with maternal inheritance was compared with paternal inheritance. Statistical analysis was performed with SPSS (version 9.0) for From the Department of Clinical Neurosciences (Drs. M. Bradley, L. Bradley, and Orrell), Royal Free and University College Medical School, University College London, London, England; and Department of Neuromuscular Diseases (Dr. de Belleroche), Imperial College School of Medicine, London, England. Supported by the Motor Neurone Disease Association, The Peel Medical Research Trust, and the Peter Samuel Royal Free Fund. Received November 18, 2004. Accepted in final form January 18, 2005. Address correspondence and reprint requests to Dr. Richard W. Orrell, Department of Clinical Neurosciences, Royal Free and University College Medical School, University College London, Rowland Hill Street, London NW3 2QG, England; r.orrell@rfc.ucl.ac.uk 1628 Copyright © 2005 by AAN Enterprises, Inc. Windows (SPSS Inc.). Direct transmissions were those between first-degree relatives (mother/father and son/daughter), and indirect transmissions were those between second-degree relatives (uncle/aunt and nephew/niece). Analyses were also performed on those members of G1 below and above the median (i.e., the lower and upper 50% of G1) to identify the statistical phenomenon of regression to the mean, where anticipation may be demonstrated in the offspring of older parents.1,2 Results. A total of 160 first-degree intergenerational transmissions were identified. Mean age at death was 55.2 years (95% CI: 54.0 to 56.4 years; n ⫽ 503) and mean disease duration was 3.5 years (CI: 2.9 to 4.1; n ⫽ 227). There was no difference (p ⬎ 0.05) in age at death between patients with SOD1 mutations (53.9 years; CI: 51.4 to 56.4; n ⫽ 114) and those with none (55.6 years; CI: 54.3 to 56.9; n ⫽ 389) or duration of disease between patients with SOD1 mutations (4.9 years; CI: 3.4 to 4.6; n ⫽ 34) and those with none (3.3 years; CI: 2.7 to 3.9, n ⫽ 193). Median age at death was compared between the two generations (table 1). For all transmissions, the Wilcoxon rank sum test demonstrated higher median age at death in G1 (55 years) than G2 (48 years) (p ⬍ 0.001). The significance was present when studying the upper (p ⬍ 0.001) and middle (p ⬍ 0.001) 50% (i.e., 25th to 75th centile) of the G1 range. When the lower 50% of G1 was studied, there was no significant difference (p ⬎ 0.05). Comparing difference in age at death between firstdegree G1 and G2 pairs for the whole sample, the 95% CI did not contain zero, indicating that age at death in G1 was significantly greater than G2 (figure). A greater deviation from zero occurred when including only the older 50% of G1, and a negative deviation with the younger 50% members of G1. Frequency of positive (anticipation) and negative intergenerational differences was compared using ␹2 test (table 2). This demonstrated the phenomenon of regression to the mean, i.e., the older the members of G1, the more significant the tendency toward anticipation. Studying all samples, ␹2 test was positive for anticipation in all groups, with the exception of first-degree intergenerational transfer in families with SOD1 mutations. When studying the younger 50% of G1, there was no anticipation. Comparing the frequency of maternal and paternal inheritance, in SOD1-positive and -negative families, ␹2 test demonstrated no difference in frequency of maternal and paternal transmissions, for both SOD1-positive and -negative families (p ⬎ 0.1). This indicates that parental sex has no effect on inheritance. Table 1 Median values of age at death for G1 and G2 in families with ALS All samples All families G2 G1 G2 G1 G2 G1 291 55 48 63 54 55 48 45 p ⬍ 0.001* 160 131 53 90 70 M/F 56 184 55 63 63 56 48 67 50 57 54 55 66 56 50 50 p ⬎ 0.05† 48 56 47 p ⬍ 0.005† 52 56 53 43 48 p ⬍ 0.001* 60 p ⬍ 0.05* 49 p ⬍ 0.01* p ⬍ 0.001* p ⬍ 0.05* 50 p ⬎ 0.05† p ⬍ 0.05* 53 43 p ⬎ 0.05 56 55 p ⬍ 0.001* p ⬍ 0.001* 48 p ⬍ 0.05* 58 43 p ⬍ 0.05† 52 58 p ⬍ 0.001* 52 42 p ⬍ 0.05* 58 48 p ⬎ 0.05† 46 55 45 49 p ⬍ 0.001* 53 66 53 53 p ⬍ 0.001* p ⬍ 0.05* 107 48 G2 p ⬎ 0.05† p ⬍ 0.005* p ⬍ 0.001* 61 57 p ⬍ 0.001* 48 57 55 62 p ⬍ 0.01* 110 p ⬍ 0.001* p ⬍ 0.001* p ⬍ 0.05* M/F/A/U SOD1 49 58 M/F SOD1 64 p ⬍ 0.005* M/F/A/U Non-SOD1 46 55 F Non-SOD1 53 p ⬍ 0.005* M All families 57 p ⬍ 0.001* p ⬍ 0.001* A/U All families Lower 50% of G1 G1 M/F All families Middle 50% of G1 n M/F/A/U All families Upper 50% of G1 45 p ⬎ 0.05 54 p ⬎ 0.05 48 45 p ⬎ 0.05 * p Wilcoxon rank sum test, where G1 was greater than G2 (p ⬍ 0.05). † G2 ⬎ G1. G1 ⫽ generation 1; G2 ⫽ generation 2; M ⫽ mother; F ⫽ father; U ⫽ uncle; A ⫽ aunt; SOD1 ⫽ copper/zinc superoxide dismutase. Discussion. The mean age at death and disease duration were similar to those in previous studies of ALS. Median age at death was lower in the second generation (p ⬍ 0.001) (see table). Mean difference in age at death between the two generations was positive (younger death in the second generation) (see figure) and the number of occasions that positive anticipation occurred was significant (p ⬍ 0.001) (see table 2). Significant differences were observed in median Figure. Intergeneration difference in age at death between generation 1 (G1) and generation 2 (G2), mean with 95% CI, for the upper, middle, and lower 50% of age of G1, and for copper/zinc superoxide dismutase (SOD1) and non-SOD1 families. age, mean difference in age, and frequency of earlier age at death between the generations. This supports the concept of anticipation in patients with ALS. Ascertainment bias is recognized in studies of anticipation.1,2 There is often incomplete information for age at onset because it may be difficult to precisely define the first feature of disease. Improvements in diagnostic capability and the increased family awareness once a parent is diagnosed, lead to offspring being diagnosed earlier than the parent. These forms of bias are avoided by using age at death. Patients with a genetic disease may not wish or may be unable to have children, leading to bias due to reproductive fitness. An individual with earlyonset disease may be less likely to have children. In a genetic line ending with early-onset patients, this will bias toward anticipation. Other studies used second-degree relatives to minimize this bias. Our analysis of second-degree relatives supported the presence of anticipation. Ascertainment bias may occur if individuals are not interviewed at the same age. Simultaneous onset in both generations attracts the attention of family members and their physicians. G2 would have an earlier onset than G1, favoring anticipation. This is compounded because any siblings of G2 who eventually develop the disease, negating the effect of perMay (1 of 2) 2005 NEUROLOGY 64 1629 Table 2 Number of occasions when the difference between G1 and G2 is positive (anticipation) and negative All samples All families n ⫹ 291 182 M/F/A/U Top 50% of G1 Middle 50% of G1 ⫺ ⫹ ⫺ ⫹ 109 118 23 100 ⬁ 18.29 p ⬍ 0.001* All families 160 101 131 81 50 56 56 p ⬍ 0.05* 9 29 41 3.41 p ⬎ 0.05 16 20 23 14.53 3.8 0.19 p ⬍ 0.001* p ⬎ .05 p ⬎ 0.05† 45 25 108 31 4 21 21.36 70 14 15 1.4 p ⬍ 0.001* 77 p ⬎ .05 12 58 22 1.36 p ⬎ 0.05† 35 38 65 5.22 ⬁ 5.75 7.15 p ⬍ 0.05* p ⬍ 0.001* p ⬍ 0.05* p ⬍ 0.01† 70 41 75 49 6 38 ⬁ 46 21 21 4.92 p ⬍ 0.001* 33 p ⬍ 0.05* 9 46 35 3.56 p ⬎ 0.05† 19 29 25 16.48 ⬁ 11.39 0.296 p ⬍ 0.001* p ⬍ 0.001* p ⬍ 0.001* p ⬎ 0.05 M/F/A/U 50 26 5.38 p ⬍ 0.01* SOD1 23 5.1 p ⬍ 0.001* 34 7.65 107 41 p ⬍ 0.05* M/F SOD1 10 16.27 34 M/F/A/U 110 46 1.51 p ⬍ 0.05* Non-SOD1 35 p ⬎ 0.05 5.71 184 p ⬎ 0.05 32 5.25 F Non-SOD1 53 p ⬍ 0.05* M 70 p ⬍ 0.001* 14 87 3.21 ⬁ p ⬍ 0.01* All families 65 p ⬍ 0.001* 7.34 90 49 ⫺ 11.02 A/U All families ⫹ p ⬍ 0.001* M/F All families 67 ⫺ 17.56 p ⬍ 0.001* 59 Bottom 50% of G1 32 19 M/F 18 9 22 11 14 11 3.38 4.00 3.78 0.36 p ⬎ 0.05 p ⬍ 0.05* p ⬎ 0.05 p ⬎ 0.05 Chi-square and p values are given, for the null hypothesis that G1 ⫽ G2. * Significant if p ⬍ 0.05. † G2 ⬎ G1. G1 ⫽ generation 1; G2 ⫽ generation 2; M ⫽ mother; F ⫽ father; U ⫽ uncle; A ⫽ aunt; SOD1 ⫽ copper/zinc superoxide dismutase. ceived anticipation, would not be detected at the time of data collection. Familial transmissions with SOD1 mutations would not be expected to demonstrate anticipation because the mode of inheritance is distinct from unstable trinucleotide repeat disorders. This suggests that apparent anticipation reported in SOD1 families3–5 may represent regression to the mean rather than anticipation per se. With no anticipation, the median value of G1 minus G2 will be close to zero, with CIs including zero. Regression to the mean is a phenomenon whereby higher values of G1 artificially increase the degree of perceived anticipation and vice versa for lower values of G1. This causes the value to be positive when selecting the oldest 50% of G1 and negative when selecting the youngest 50% of G1. The phenomenon 1630 NEUROLOGY 64 May (1 of 2) 2005 was observed when we used this method to analyze the data (see figure). This suggests that the initial perception of anticipation is probably a result of regression to the mean. The possibility of anticipation in a small subgroup of families with ALS cannot be completely excluded. We did not demonstrate increased maternal inheritance in familial ALS. The lack of sex effect on inheritance occurs in SOD1-positive and -negative families and does not support a major contribution of transmission of mitochondrial DNA mutations to inheritance. In conclusion, there is an initial appearance of significant anticipation in age at death between generations in familial ALS. The absence of anticipation in the offspring of the younger half of the patient population studied raises the possibility of bias, which is recognized in all anticipation studies. The recognition of apparent, but not true, anticipation, may be important in counseling families and affected individuals with ALS. References 1. Ashizawa T, Conneally PM. Repeats may not be everything in anticipation. Neurology 1999;53:1164 –1165. 2. McInnis MG. Anticipation: an old idea in new genes. Am J Hum Genet 1996;59:973–979. 3. Iwai K, Yamamoto M, Yoshihara T, Sobue G. Anticipation in familial amyotrophic lateral sclerosis with SOD1-G93S mutation. J Neurol Neurosurg Psychiatry 2002;72:819 – 820. 4. Aoki M, Abe K, Houi K, et al. Variance of age at onset in a Japanese family with amyotrophic lateral sclerosis associated with a novel Cu/Zn superoxide dismutase mutation. Ann Neurol 1995;37:676 – 679. 5. Ceroni M, Malaspina A, Poloni TE, et al. Clustering of ALS patients in central Italy due to the occurrence of the L84F SOD1 gene mutation. Neurology 1999;53:1064 –1071. 6. Leone M, De Angelis MS, Giordano M, Mutani R. Influence of ancestral gender on transmission of familial amyotrophic lateral sclerosis. Lancet 1994;344:1639. 7. Orrell RW, King AW, de Belleroche JS. Parental influence on inheritance of familial amyotrophic lateral sclerosis. Lancet 1995;345:391– 392. 8. Watanabe M, Abe K, Aoki M, et al. Mitotic and meiotic stability of the CAG repeat in the X-linked spinal and bulbar muscular atrophy gene. Clin Genet 1996;50:133–137. 9. Orrell RW, Schapira AH. Mitochondria and amyotrophic lateral sclerosis. Int Rev Neurobiol 2002;53:411– 426. 10. Orrell RW, Habgood JJ, Gardiner I, et al. Clinical and functional investigation of 10 missense mutations and a novel frameshift insertion mutation of the gene for copper-zinc superoxide dismutase in UK families with amyotrophic lateral sclerosis. Neurology 1997;48:746 –751. NeuroImages Figure. CT brain scan showed a left subdural hematoma with 1.5 cm midline shift (panel 7/24). After emergency platelet transfusion, the subdural hematoma was evacuated. Brain CT the next day showed resolution of the shift (panel 7/25). Brain MRI a week later showed a cerebral peduncle lesion in the T2-weighted sequence (panel 7/31). There was no history of head trauma. Kernohan notch lesion after spinal tap Harry Openshaw, MD, Duarte, CA A 32-year-old woman with acute lymphocytic leukemia had abrupt onset of headache, decreased sensorium, and dilated left pupil. Lumbar punctures with intrathecal methotrexate had been done 4 and 7 days before her collapse. When she awakened from emergency evacuation of a subdural hematoma, there was ipsilateral hemiplegia and MRI showed a cerebral peduncle lesion (figure). Now, 32 months later, hemiparesis persists. Subdural hematomas are rare complications of lumbar puncture, even in patients receiving intrathecal chemotherapy for leukemia.1 The probable mechanism involves intracranial hypotension, subdural hygroma formation, stretching of the bridging veins, and bleed- Address correspondence and reprint requests to Dr. Harry Openshaw, Department of Neurology, City of Hope National Medical Center, 1500 East Duarte Road, Duarte, CA 91010-3000; e-mail: hopenshaw@coh.org ing in the context of thrombocytopenia. Kernohan notch hemiplegia may occur during transtentorial herniation. The lesion has been well explained anatomically,2,3 but MRI demonstration and long-term patient survival are unusual.4 Copyright © 2005 by AAN Enterprises, Inc. 1. Jourdan E, Dombret H, Glaisner S, Miclea JM, Castaigne S, Degos L. Unexpected high incidence of intracranial subdural haematoma during intensive chemotherapy for acute myeloid leukaemia with a monoblastic component. Br J Haematol 1995;89:527–530. 2. Kernohan JW, Woltman HW. Incisura of the crus due to contralateral brain tumor. Arch Neurol Psychiatry 1929;21:274 –287. 3. Adler E, Milhorat T. The tentorial notch: anatomical variation, morphometric analysis, and classification in 100 human autopsy cases. J Neurosurg 2002;96:1103–1112. 4. Binder D, Lyon R, Manley G. Transcranial motor evoked potential recording in a case of Kernohan’s notch syndrome: case report. Neurosurgery 2004;4:999 –1002. May (1 of 2) 2005 NEUROLOGY 64 1631