Re: Common conditions associated with hereditary haemochromatosis genetic variants: cohort study in UK Biobank
Rapid responses to our paper on HFE p.C282Y associated morbidity in UK Biobank[1] have raised issues that require clarification.
A. Claimed contradictory results to previous community studies
Professor Worwood and Dr McCune highlight their studies of Welsh blood donors and relatives [2-4], which reported similar p.C282Y mutation prevalence to UK Biobank. Their studies also found no excess clinical morbidity in p.C282Y heterozygotes (again similar to UK Biobank): we apologise for not citing these studies.
Worwood and McCune go on to argue that their findings of no excess clinical disease in p.C282Y homozygotes suggest that our UK Biobank findings are invalid. The Welsh studies analysed disease associations in n=72 p.C282Y male and female homozygotes (mean age <40 years) and n=59 p.C282Y homozygous relatives (mean age 52). Samples of this size have low power to detect the expected associations in such relatively young groups, especially if pooling data from men and women. In the eMERGE study across 7 US hospital systems[5], only four of 95 p.C282Y homozygotes with relevant data were diagnosed with haemochromatosis before age 50: most diagnoses were made after the age of 50, with several at age 80 plus. Our paper noted the similar limitations of the Beutler et al[6] study (mean age 50 with n=56 p.C282Y homozygous men), which had only 19% power (at p=0.05) to detect the Biobank p.C282Y male homozygote all ages association with diabetes.
Despite lack of power, these previous studies actually produce broadly similar results to those in younger Biobank participants, which also showed no excess liver disease or diabetes (age 40 to 49 years, n=290 men and n=339 women). An association was present with osteoarthritis in men only, although prevalences in this age-group were low.
It should be noted that the largest previous community study (HIERS) did find significant associations between p.C282Y homozygotes[7] and liver disease in men, and higher prevalence of chronic fatigue and metacarpophalangeal joint swelling in p.C282Y homozygotes with higher serum ferritin levels.
Space does not allow discussion of the many clinical and family studies which support the UK Biobank results of substantial excess morbidity at older ages (50 to 70 at baseline) associated with p.C282Y homozygosity, especially in men. As with other risk factors including high blood pressure or cholesterol levels, lack of association with related disease at mean ages <53 tells us little about the later life excess morbidity, and the true burden of p.C282Y homozygous status would be best assessed on a lifetime basis.
B. Inclusion of (likely misdiagnoses) of rheumatoid arthritis,
Drs Kiely as well as Worwood and McCune question our inclusion of rheumatoid arthritis diagnoses. However, the similarity in presentation of some Hereditary Haemochromatosis arthroses with rheumatoid arthritis is well documented[8]. Our paper noted that “misdiagnoses of iron related pathology has been reported, for example with the iron deposition related arthrosis misdiagnosed as osteoarthritis or rheumatoid”. We apologise that these likely misdiagnosis were insufficiently clear, but do not accept that this result should have been hidden from patients. We hope this finding will contribute to more appropriate treatment for this group in the future.
C. Diabetes and body mass index
Worwood and McCune argue that p.C282Y homozygote associations we reported with diabetes mellitus and liver disease may be explained by obesity. In Biobank, p.C282Y homozygote status was associated with substantially lower rates of obesity (e.g. males OR 0.81, 95% CI 0.69 to 0.95, p=0.01; BMI >30 vs Normal 18.5 to 25), compared to no p.C282Y mutations. When BMI is adjusted for, the p.C282Y homozygous association with prevalent diabetes became stronger (e.g. males OR 1.65, 95% CI 1.26 to 2.15, p=0.0002). Adjustment for BMI had no effect on liver disease associations.
D. Alternative Estimates of associated morbidity
Worwood and McCune speculate that HH diagnoses may be dominated by people with high iron levels and no related morbidity. In UK Biobank 40 to 70 year olds, 48% of the 281 male p.C282Y homozygotes with diagnosed haemochromatosis also had liver disease, diabetes or arthritis diagnoses by the end of follow-up, compared to 15.7% in wild-type males (n=175,539): therefore 32 in 100 p.C282Y homozygotes (95% CI 26.5 to 38.2, p<0.00001) had excess diagnoses. As our paper noted, this is likely a substantial underestimate, as diagnoses after baseline were available from inpatient hospital records only, while most diagnoses of diabetes and arthritis are made in primary care or outpatient settings.
Ignoring diagnoses of haemochromatosis (as requested), e.g. male p.C282Y homozygotes still had substantially more diagnosed liver disease, arthritis or diabetes by end follow-up (mean age only 63 years: 27% in p.C282Y homozygotes versus 16% in wild type: difference 11 (95% CI 9 to 14) additional homozygous patients with diagnosed morbidity per hundred, p<0.00001). Excess morbidity in 40 to 49 year old males by the same measure was small (13% versus 6%, excess 6 CI 3 to 10), and would be very challenging to detect without a large sample.
E. Ascertainment bias and prevalent diagnosis of haemochromatosis
Worwood and McCune suggest that ascertainment bias and self-reported illness with diagnosed haemochromatosis or in family members could explain away our finding. Our paper reported incident associations from hospital inpatient diagnoses only, and found excess morbidity in p.C282Y homozygotes, with a marked gender difference. We also reported such excess morbidity for ‘hard endpoints’ including hip replacements, incident liver cancers and liver cancer deaths. In a sister paper[9] we analysed measured muscle (grip) strength in 60 to 70 year olds. Low measured strengths were more common in male p.C282Y homozygotes (24% versus 15% in wild type: OR 1.69: CI 1.39 to 2.05). Overall therefore it seems implausible that ascertainment bias could explain away all the excess morbidity observed in p.C282Y homozygotes in UK Biobank. It should also be noted that haemochromatosis diagnoses and family histories are likely to lead to early venesection, thus biasing penetrance estimates toward the null.
We look forward to the imminent release of liver enzyme and other data in the UK Biobank cohort, to strengthen the evidence further.
References
1. Pilling LC, Tamosauskaite J, Jones G, et al. Common conditions associated with hereditary haemochromatosis genetic variants: cohort study in UK Biobank. BMJ. 2019;364:k5222. doi:10.1136/bmj.k5222.
2. Jackson HA, Carter K, Darke C, Guttridge MG, Ravine D, Hutton RD, Napier JA and Worwood M. HFE mutations, iron deficiency and overload in 10 500 blood donors. Br J Haematol 2001; 114: 474–484.
3. McCune CA, Al Jader LN, May A, Hayes SL, Jackson HA and Worwood M. Hereditary haemochromatosis: only 1% of adult HFE C282Y homozygotes in South Wales have a clinical diagnosis of iron overload. Hum Genet 2002; 111: 538–543
4. C A McCune, D Ravine, K Carter, H A Jackson, D Hutton, J Hedderich, M Krawczak, M Worwood. Iron loading and morbidity among relatives of HFE C282Y homozygotes identified either by population genetic testing or presenting as patients. Gut; 2006; 55: 554–562.
5. Gallego CJ, Burt A, Sundaresan AS, et al. Penetrance of Hemochromatosis in HFE Genotypes Resulting in p.Cys282Tyr and p.[Cys282Tyr];[His63Asp] in the eMERGE Network. Am J Hum Genet. 2015;97(4):512-520. doi:10.1016/j.ajhg.2015.08.008.
6. Beutler E, Felitti VJ, Koziol JA, Ho NJ, Gelbart T. Penetrance of 845G → A (C282Y) HFE hereditary haemochromatosis mutation in the USA. Lancet. 2002;359(9302):211-218. doi:10.1016/S0140-6736(02)07447-0.
7. McLaren GD, Gordeuk VR. Hereditary hemochromatosis: insights from the Hemochromatosis and Iron Overload Screening (HEIRS) Study. Hematol Am Soc Hematol Educ Progr. 2009;6:195-206. doi:10.1182/asheducation-2009.1.195.
8. Husar-Memmer E, Stadlmayr A, Datz C, Zwerina J. HFE-related hemochromatosis: an update for the rheumatologist. Curr Rheumatol Rep. 2014;16(1):393. doi:10.1007/s11926-013-0393-4.
9. Tamosauskaite J, Atkins JL, Pilling LC, et al. Hereditary Hemochromatosis Associations with Frailty, Sarcopenia and Chronic Pain: Evidence from 200,975 Older UK Biobank Participants. J Gerontol A Biol Sci Med Sci. 2019;In press. doi:10.1093/gerona/gly270.
Rapid Response:
Re: Common conditions associated with hereditary haemochromatosis genetic variants: cohort study in UK Biobank
Rapid responses to our paper on HFE p.C282Y associated morbidity in UK Biobank[1] have raised issues that require clarification.
A. Claimed contradictory results to previous community studies
Professor Worwood and Dr McCune highlight their studies of Welsh blood donors and relatives [2-4], which reported similar p.C282Y mutation prevalence to UK Biobank. Their studies also found no excess clinical morbidity in p.C282Y heterozygotes (again similar to UK Biobank): we apologise for not citing these studies.
Worwood and McCune go on to argue that their findings of no excess clinical disease in p.C282Y homozygotes suggest that our UK Biobank findings are invalid. The Welsh studies analysed disease associations in n=72 p.C282Y male and female homozygotes (mean age <40 years) and n=59 p.C282Y homozygous relatives (mean age 52). Samples of this size have low power to detect the expected associations in such relatively young groups, especially if pooling data from men and women. In the eMERGE study across 7 US hospital systems[5], only four of 95 p.C282Y homozygotes with relevant data were diagnosed with haemochromatosis before age 50: most diagnoses were made after the age of 50, with several at age 80 plus. Our paper noted the similar limitations of the Beutler et al[6] study (mean age 50 with n=56 p.C282Y homozygous men), which had only 19% power (at p=0.05) to detect the Biobank p.C282Y male homozygote all ages association with diabetes.
Despite lack of power, these previous studies actually produce broadly similar results to those in younger Biobank participants, which also showed no excess liver disease or diabetes (age 40 to 49 years, n=290 men and n=339 women). An association was present with osteoarthritis in men only, although prevalences in this age-group were low.
It should be noted that the largest previous community study (HIERS) did find significant associations between p.C282Y homozygotes[7] and liver disease in men, and higher prevalence of chronic fatigue and metacarpophalangeal joint swelling in p.C282Y homozygotes with higher serum ferritin levels.
Space does not allow discussion of the many clinical and family studies which support the UK Biobank results of substantial excess morbidity at older ages (50 to 70 at baseline) associated with p.C282Y homozygosity, especially in men. As with other risk factors including high blood pressure or cholesterol levels, lack of association with related disease at mean ages <53 tells us little about the later life excess morbidity, and the true burden of p.C282Y homozygous status would be best assessed on a lifetime basis.
B. Inclusion of (likely misdiagnoses) of rheumatoid arthritis,
Drs Kiely as well as Worwood and McCune question our inclusion of rheumatoid arthritis diagnoses. However, the similarity in presentation of some Hereditary Haemochromatosis arthroses with rheumatoid arthritis is well documented[8]. Our paper noted that “misdiagnoses of iron related pathology has been reported, for example with the iron deposition related arthrosis misdiagnosed as osteoarthritis or rheumatoid”. We apologise that these likely misdiagnosis were insufficiently clear, but do not accept that this result should have been hidden from patients. We hope this finding will contribute to more appropriate treatment for this group in the future.
C. Diabetes and body mass index
Worwood and McCune argue that p.C282Y homozygote associations we reported with diabetes mellitus and liver disease may be explained by obesity. In Biobank, p.C282Y homozygote status was associated with substantially lower rates of obesity (e.g. males OR 0.81, 95% CI 0.69 to 0.95, p=0.01; BMI >30 vs Normal 18.5 to 25), compared to no p.C282Y mutations. When BMI is adjusted for, the p.C282Y homozygous association with prevalent diabetes became stronger (e.g. males OR 1.65, 95% CI 1.26 to 2.15, p=0.0002). Adjustment for BMI had no effect on liver disease associations.
D. Alternative Estimates of associated morbidity
Worwood and McCune speculate that HH diagnoses may be dominated by people with high iron levels and no related morbidity. In UK Biobank 40 to 70 year olds, 48% of the 281 male p.C282Y homozygotes with diagnosed haemochromatosis also had liver disease, diabetes or arthritis diagnoses by the end of follow-up, compared to 15.7% in wild-type males (n=175,539): therefore 32 in 100 p.C282Y homozygotes (95% CI 26.5 to 38.2, p<0.00001) had excess diagnoses. As our paper noted, this is likely a substantial underestimate, as diagnoses after baseline were available from inpatient hospital records only, while most diagnoses of diabetes and arthritis are made in primary care or outpatient settings.
Ignoring diagnoses of haemochromatosis (as requested), e.g. male p.C282Y homozygotes still had substantially more diagnosed liver disease, arthritis or diabetes by end follow-up (mean age only 63 years: 27% in p.C282Y homozygotes versus 16% in wild type: difference 11 (95% CI 9 to 14) additional homozygous patients with diagnosed morbidity per hundred, p<0.00001). Excess morbidity in 40 to 49 year old males by the same measure was small (13% versus 6%, excess 6 CI 3 to 10), and would be very challenging to detect without a large sample.
E. Ascertainment bias and prevalent diagnosis of haemochromatosis
Worwood and McCune suggest that ascertainment bias and self-reported illness with diagnosed haemochromatosis or in family members could explain away our finding. Our paper reported incident associations from hospital inpatient diagnoses only, and found excess morbidity in p.C282Y homozygotes, with a marked gender difference. We also reported such excess morbidity for ‘hard endpoints’ including hip replacements, incident liver cancers and liver cancer deaths. In a sister paper[9] we analysed measured muscle (grip) strength in 60 to 70 year olds. Low measured strengths were more common in male p.C282Y homozygotes (24% versus 15% in wild type: OR 1.69: CI 1.39 to 2.05). Overall therefore it seems implausible that ascertainment bias could explain away all the excess morbidity observed in p.C282Y homozygotes in UK Biobank. It should also be noted that haemochromatosis diagnoses and family histories are likely to lead to early venesection, thus biasing penetrance estimates toward the null.
We look forward to the imminent release of liver enzyme and other data in the UK Biobank cohort, to strengthen the evidence further.
References
1. Pilling LC, Tamosauskaite J, Jones G, et al. Common conditions associated with hereditary haemochromatosis genetic variants: cohort study in UK Biobank. BMJ. 2019;364:k5222. doi:10.1136/bmj.k5222.
2. Jackson HA, Carter K, Darke C, Guttridge MG, Ravine D, Hutton RD, Napier JA and Worwood M. HFE mutations, iron deficiency and overload in 10 500 blood donors. Br J Haematol 2001; 114: 474–484.
3. McCune CA, Al Jader LN, May A, Hayes SL, Jackson HA and Worwood M. Hereditary haemochromatosis: only 1% of adult HFE C282Y homozygotes in South Wales have a clinical diagnosis of iron overload. Hum Genet 2002; 111: 538–543
4. C A McCune, D Ravine, K Carter, H A Jackson, D Hutton, J Hedderich, M Krawczak, M Worwood. Iron loading and morbidity among relatives of HFE C282Y homozygotes identified either by population genetic testing or presenting as patients. Gut; 2006; 55: 554–562.
5. Gallego CJ, Burt A, Sundaresan AS, et al. Penetrance of Hemochromatosis in HFE Genotypes Resulting in p.Cys282Tyr and p.[Cys282Tyr];[His63Asp] in the eMERGE Network. Am J Hum Genet. 2015;97(4):512-520. doi:10.1016/j.ajhg.2015.08.008.
6. Beutler E, Felitti VJ, Koziol JA, Ho NJ, Gelbart T. Penetrance of 845G → A (C282Y) HFE hereditary haemochromatosis mutation in the USA. Lancet. 2002;359(9302):211-218. doi:10.1016/S0140-6736(02)07447-0.
7. McLaren GD, Gordeuk VR. Hereditary hemochromatosis: insights from the Hemochromatosis and Iron Overload Screening (HEIRS) Study. Hematol Am Soc Hematol Educ Progr. 2009;6:195-206. doi:10.1182/asheducation-2009.1.195.
8. Husar-Memmer E, Stadlmayr A, Datz C, Zwerina J. HFE-related hemochromatosis: an update for the rheumatologist. Curr Rheumatol Rep. 2014;16(1):393. doi:10.1007/s11926-013-0393-4.
9. Tamosauskaite J, Atkins JL, Pilling LC, et al. Hereditary Hemochromatosis Associations with Frailty, Sarcopenia and Chronic Pain: Evidence from 200,975 Older UK Biobank Participants. J Gerontol A Biol Sci Med Sci. 2019;In press. doi:10.1093/gerona/gly270.
Competing interests: No competing interests