OBJECTIVE: Alaska Native and other circumpolar indigenous populations have historically experienced high infant mortality rates, for unknown reasons. Through routine newborn screening, Alaskan and Canadian indigenous infants have been found to have a high frequency of a single sequence variant (c.1436C→T) in the gene coding for carnitine palmitoyltransferase type 1A (CPT1A). We sought to determine whether these 2 findings were related.
METHODS: As part of a quality control exercise at the Alaskan Newborn Metabolic Screening Program, we conducted genotyping for 616 consecutively born, Alaska Native infants and reviewed their medical records. We conducted an ecological analysis comparing Census area–level variant CPT1A allele frequency and historical Alaska Native infant, postneonatal, and neonatal mortality rates.
RESULTS: Infant death was identified for 5 of 152 infants homozygous for the c.1436C→T sequence variant (33 deaths per 1000 live births), 2 of 219 heterozygous infants (9 deaths per 1000 live births), and 0 of 245 infants carrying no copies of the variant allele (χ2 = 9.2; P = .01). All 7 cases of infant death had some evidence of an infectious process at the time of death, including 5 with respiratory infections. Census areas with the highest frequency of the variant allele had the highest historical infant, postneonatal, and neonatal mortality rates.
CONCLUSIONS: Our data provide preliminary evidence that a highly prevalent CPT1A variant found among Alaska Native and other indigenous circumpolar populations may help explain historically high infant mortality rates. Larger definitive studies are needed.
- Alaska Native
- carnitine palmitoyltransferase deficiency
- fatty acid oxidation disorder
- infant death
- metabolic disorders
- newborn screening
- sudden infant death syndrome
WHAT'S KNOWN ON THIS SUBJECT:
For unknown reasons circumpolar indigenous populations such as Alaska Native people have historically experienced high infant mortality rates.
WHAT THIS STUDY ADDS:
Infants homozygous for the c.1436C→T variant in the gene coding for carnitine palmitoyltransferase type 1A (CPT1A) had higher infant mortality rates, primarily attributable to infectious diseases, than did other infants. Regions with the highest infant mortality rates had the highest gene frequencies.
Historically and for unknown reasons, Alaska Native and American Indian infants have had increased risk of death, especially during the postneonatal period, primarily attributable to sudden infant death syndrome (SIDS) and infections.1,–,5 Potential risk factors include poverty, poor maternal education, parental drug use, family disintegration, household crowding, lack of clean water and septic services, and increased risk of infectious diseases of infancy, such as Haemophilus influenzae type b, Streptococcus pneumoniae, and respiratory syncytial virus.1,2,6,–,9
Recently, great strides have been made in reducing postneonatal mortality rates among this population in some areas of the Pacific Northwest.10,11 However, this has not occurred in Alaska in the past 10 years, where Alaska Native postneonatal mortality rates have remained unchanged and high. This is particularly true for infants residing in Alaska's northern and western regions, which are populated mainly by Yupik and Inuit people (who constitute subpopulations of the broader Alaska Native population). The summary infant mortality rates in these two regions during 2001–2005 for all races were 12.5 and 9.6 deaths per 1000 live births, respectively, compared to a range of 5.7 to 6.4 deaths per 1000 live births in Alaska's other 4 regions.3
Previous studies documented associations between fatty acid oxidation disorders and sudden, unexpected, infant death.12,–,14 These disorders are rare and were not known previously to occur at high frequency in indigenous populations. Recent studies, however, reported a high prevalence among Canadian Inuit populations of a single sequence variant (c.1436C→T) in the gene coding for carnitine palmitoyltransferase type 1A (CPT1A).15,–,17 CPT1A deficiency is a rare autosomal recessive disorder of fatty acid oxidation that impairs fasting ketogenesis by the liver, but the clinical significance of homozygosity for this variant, which results in only partial loss of CPT1A activity, is not known.
Expanded newborn screening with tandem mass spectrometry began in Alaska in 2003, and fatty acid oxidation disorders, including CPT1A deficiency, were part of the routine panel offered by the Oregon Public Health Laboratory, which performed newborn screening tests for Alaska. Testing quickly identified a high prevalence of CPT1A deficiency attributable to the c.1436C→T variant among Alaska Native infants. The affected infants were predominantly from the western and northern regions of Alaska, where SIDS and postneonatal mortality rates are highest.
As part of a quality control assessment to address concerns regarding the sensitivity of tandem mass spectrometry–based newborn screening to identify infants with CPT1A deficiency attributable to the c.1436C→T variant, 2499 consecutive newborn bloodspot screening cards from Alaskan infants were genotyped for the presence of the 1436C or 1436T allele of CPT1A. The diagnosis of CPT1A deficiency via MS/MS is based on the presence of an elevated ratio of free carnitine (C0) to the sum of C16 and C18 acylcarnitines (C0/C16 + C18). During the period of this analysis, a C0/C16 + C18 value of >130 was considered positive for CPT1A deficiency. Compared with genotyping as the standard method, tandem mass spectrometry had a sensitivity of 0% (0 of 166), on the basis of a standard cutoff value of 130. Genotype analysis also demonstrated the almost-exclusive occurrence (99%; all except 1 infant) of the variant allele among infants with ≥1 parent reported to be Alaska Native.18 As a group, 26% of all Alaska Native infants (165 of 633 infants) in the evaluated sample were homozygous for the variant allele; among those from the western and northern regions, however, >50% were homozygous. On the basis of these initial results, we aimed to determine whether CPT1A genotype predicted infant death among evaluated infants and whether an ecological association existed between Census area–level CPT1A allele frequency among evaluated infants and historical Alaska Native infant, postneonatal, and neonatal mortality rates.
Evidence Line 1
As a first line of evidence, we calculated infant (0–364 days), postneonatal (28–364 days), and neonatal (0–27 days) mortality rates according to genotype. This was accomplished by linking genotyping results to birth and death certificates provided by the Alaska Bureau of Vital Statistics. Of the 633 Alaska Native infants with genotype data, 616 could be linked with a birth certificate. Because of the small Alaska Native birth cohort, linkage was complete on the basis of infant name, date of birth, and residence. Infant birth weight and gestational age and maternal age, education, and history of prenatal alcohol or tobacco use all have been historically associated with infant death in Alaska,1,–,3,19 but these factors did not differ between infants who did and did not die (Table 1). Residence in the western or northern regions did differ; therefore, results limited to this area also are presented. The Alaska Division of Public Health Maternal-Infant Mortality Review was used to access autopsy results for children in the study cohort who died.
Evidence Line 2
As a second line of evidence, we conducted an ecological analysis to evaluate the association between the current frequency of the variant allele within geographic Census areas (Alaska does not have counties) and historic Alaska Native infant, postneonatal, and neonatal mortality rates within those areas. On the basis of the genotyping results for the cohort of 616 evaluated infants, the variant allele frequency in a particular Census area was calculated as one-half the number of heterozygous children in the Census area plus the number of children homozygous for the variant divided by the total number of tested children. Although the c.1436C→T allele frequency was used as the primary independent variable of interest, we repeated all analyses using the proportions of homozygous infants and infants either homozygous or heterozygous for the c.1436C→T variant. Because these did not change any of the conclusions, only the allele frequency associations are presented.
Census area mortality rates were determined from a linked birth-death database for 1980–2007. A recent analysis found that maternal substance use, lack of a father's name on the birth certificate, and maternal education were the strongest evaluated predictors of infant death in Alaska.1 Consequently, we used the vital records database to determine Census area–level values for the proportions of Alaska Native infants born to unmarried women with no father's name on the birth certificate and to women with documented prenatal tobacco or alcohol use. We used data provided by the Alaska Department of Labor, as described previously,19 to determine the proportion of adults ≥25 years of age with <12 years of formal education.
For each of the 3 mortality outcomes, we constructed separate linear regression models with data for the 19 (of 27 total) Census areas in Alaska with ≥5 infants in the study population and ≥500 Alaska Native infant births in 1980–2007. All models contained 4 independent variables, that is, the variable of interest (CPT1A allele frequency) and all 3 potential confounding variables. For all independent variables and models, the variance inflation factor was <4.0, indicating modest colinearity. Only simple linear models were constructed. Although nonlinear models also might have described the data, the small numbers of deaths and tested infants in some Census areas and the resultant uncertainty surrounding estimates for both independent and dependent variables made it impossible to compare different model structures adequately.
The original evaluation providing genotype data for infants served by the Alaska Newborn Screening Program was undertaken as a quality control exercise to evaluate the sensitivity of tandem mass spectrometry–based newborn screening for CPT1A deficiency attributable to the c.1436C→T variant. Newborn bloodspots, birth certificates, and other sources of public health data used in that study and the current evaluation are under the legal jurisdiction of the Alaska Division of Public Health, which is legally mandated to conduct newborn metabolic screening for Alaskan infants. In Alaska, routine screening includes evaluation for carnitine palmitoyltransferase deficiency, and the responsibility of the Alaska Division of Public Health regarding this condition is the same as for all other screened conditions. Consequently, the analysis of these data for public health purposes by employees of the Alaska Division of Public Health is considered exempt from the need for institutional review board review.
The Alaska Newborn Screening program has an advisory panel that reviews policies and procedures, including the current evaluation, and has representation from the Alaska Native community. In addition, all work performed with respect to this evaluation and other work on CPT1A in Alaska is performed in full collaboration with the Alaska Native community, including the Alaska Native Tribal Health Consortium, which supports and in some cases actively participates in evaluation and research regarding CPT1A deficiency. Results of the current evaluation were presented to Alaska Native Health leaders, who have voiced strong support for additional, more-definitive evaluation of the association between the occurrence of the c.1436C→T sequence variant and Alaska Native infant deaths.
Testing for the c.1436C→T sequence variant in DNA extracted from newborn screening cards at the Oregon Health and Science University was conducted by using anonymous identifiers. The protocol was reviewed by the Oregon Health and Science University institutional review board and was granted an exemption (institutional review board exemption 3561).
Mortality Rates According to Genotype
Within our study population of 616 Alaska Native infants, there were 7 deaths (ie, mortality rate of 11.4 deaths per 1000 live births). Infant death was identified for 5 of 152 infants homozygous for the c.1436C→T sequence variant (33 deaths per 1000 live births), 2 of 219 heterozygous infants (9 deaths per 1000 live births), and 0 of 245 infants carrying no copies of the variant allele (χ2 = 9.2; P = .01). Compared with noncarriers and heterozygotes, homozygous infants had an infant mortality risk of 7.6 (95% confidence interval [CI]: 1.5–38.9; Fisher's exact test, P = .01). When analyses were limited to infants from the western and northern regions, infant death was identified for 5 of 121 homozygous infants (41 deaths per 1000 live births) and 1 of 122 noncarriers and heterozygotes (8 deaths per 1000 live births) (mortality risk: 5.0 [95% CI: 0.6–42.6]; Fisher's exact test, P = .12).
Of the 5 deaths involving homozygous infants, 3 occurred out of hospital during the postneonatal period, including 1 attributable to spontaneous peritonitis, 1 attributable to SIDS, and 1 attributable to homicide. The latter 2 infants demonstrated inflammatory pulmonary infiltrates at autopsy, but these infiltrates were thought to be insufficient to explain death. The other 2 homozygous infants died during the neonatal period, 1 with a congenital diaphragmatic hernia, a large ventricular septal defect, pulmonary hypertension, and possible sepsis and 1 with extreme prematurity who died at 4 weeks of age, with pneumonia and pleural effusions listed on the death certificate as possible causes of death. The 2 deaths involving heterozygous infants both occurred during the postneonatal period, with 1 attributable to interstitial pneumonitis and 1 attributable to pneumonia. All 7 cases of infant death in our study population had some evidence of an infectious process at the time of death, including 5 with respiratory infections, although none had a specific causal organism identified.
Ecological Analysis of Census Area–Level Mortality Rates According to Allele Frequency
For the 19 evaluated Census areas, the variant CPT1A allele frequency among Alaska Native infants included in our study varied from 0.10 to 0.85 (mean: 0.41; median: 0.31). For Alaska Native infants, the Census area level infant mortality rate during 1980–2007 varied from 3.8 to 21.4 deaths per 1000 live births (mean: 11.8 deaths per 1000 live births; median: 10.7 deaths per 1000 live births), postneonatal mortality rates from 3.3 to 11.8 deaths per 1000 live births (mean: 6.9 deaths per 1000 live births; median: 6.7 deaths per 1000 live births), and neonatal mortality rates from 0.5 to 9.6 deaths per 1000 live births (mean: 4.8 deaths per 1000 live births; median: 4.4 deaths per 1000 live births).
In ecological analyses at the Census area level, linear regression models demonstrated that the CPT1A variant allele frequency predicted infant and postneonatal mortality rates (Table 2 and Fig 1). For each 1% increase in allele frequency, infant mortality rates increased by 0.124 death per 1000 live births. Stratified analysis to assess effect modification by variables entered into the models was not possible because of small numbers.
The current analysis identified an increased risk of death among infants who were homozygous for the c.1436C→T variant of CPT1A. An ecological analysis demonstrated an association between Census area-level variant CPT1A allele frequency and historical infant, postneonatal, and neonatal mortality rates. Because the sample size for the direct evaluation was small and because the ecological evaluation was indirect, however, we view our results as exploratory.
Although an association between inherited disorders of fatty acid oxidation and sudden, unexpected, infant death is well known,12,–,14 the CPT1A variant prevalent in Alaska and other circumpolar areas results in only partial (∼80%) loss of catalytic activity, on the basis of which a relatively mild clinical phenotype might be predicted.15,20 Despite such predictions, severe consequences have been reported for infants and children homozygous for this variant.15,16 We also observed biochemical evidence of impaired fatty acid oxidation and reduced ketone production in affected children, including 1 child who presented with recurrent episodes of hypoglycemia and seizures (D.M.K. and M.B.G., unpublished data, 2010). Although under most circumstances homozygosity for the c.1436C→T variant may have little significant clinical impact, in some cases, such as in the presence of significant stress, it may increase the risk of death.
If an association between infant death and the CPT1A variant is confirmed, it would strongly suggest that this detrimental effect must be balanced by a selective advantage that led to the high gene frequency within affected populations. The c.1436C→T variant results in substitution of leucine for proline at amino acid 479 in CPT1A (p.P479L), which not only reduces catalytic activity but also affects regulation, reducing sensitivity to the physiologic signals that normally suppress fatty acid oxidation when sufficient glucose is available. The traditional diet in northern and western Alaska is composed primarily of marine mammals, fish and game, fruits, and various leafy plants, which as a whole provide 80% to 85% of energy as fat and 15% to 20% as protein, with very little carbohydrate. The ability to sustain a perpetual state of ketogenesis resulting from the effect of the variant on CPT1A regulation may be beneficial when such a diet is consumed. The populations in which the c.1436C→T variant is most prevalent also live in the circumpolar region, where winters are long, dark, and extremely cold. The variant might have increased the ability to survive in these conditions through effects on thermogenesis or slowing of the rate of oxidation of both stored and dietary fats. Alternatively, the high prevalence may be attributable merely to tight linkage to an unrelated gene that confers a selective advantage.
All infants who died in our study population had some evidence of an infectious process, including 5 with respiratory infections. Historically, the western and northern regions of Alaska, where CPT1A allele frequency is highest, have had higher cause-specific mortality rates for most causes of infant death. However, the increase in cause-specific risk for Alaska Native infants is most pronounced for infectious causes of death, specifically those attributable to respiratory infections. This is demonstrated by data from the Alaska Maternal-Infant Mortality Review (Table 3), which evaluates all infant deaths and uses an expert panel, composed primarily of physicians, to review all available data to assign cause of death. Together with the results from the current study, this raises the possibility that the stress of infection combines with CPT1A deficiency to increase the risk of infant death, possibly through severe hypoglycemia.
On the basis of these findings, the Alaska Division of Public Health, the Alaska Native Tribal Health Consortium, and Oregon Health and Science University have partnered to develop educational materials, including videos and pamphlets, for Alaska Native families. In addition, an extensive educational effort has been directed toward health care providers, including village-level community health aides. The educational message emphasizes regular feeding, with age-adjusted fasting limitations; increased attention to energy and fluid intake during periods of acute stress, such as during infections; and prompt medical care for acute illnesses or other situations affecting energy intake, particularly during infancy. Because of the expense associated with genotyping, the uncertain clinical consequences of the c.1436C→T variant of CPT1A, and the poor sensitivity of tandem mass spectrometry, educational messages are targeted to the Alaska Native community as a whole.
The primary strength of the current study was that we used results based on genotyping (versus tandem mass spectrometry), with a nearly complete link to vital records data. Nevertheless, our analysis had noteworthy limitations. By far the greatest limitation was the limited number of genotyping results available to us for analysis; this affected both the direct and ecological evaluations. We currently are searching for funding to conduct an analysis of 3 full years of births, which would address this limitation. The ecological analysis also was limited by the small number of births in some Census areas, despite the inclusion of 28 years of data, which led to unstable estimates of mortality rates. A more-robust direct analysis would eliminate the need for an ecological analysis.
For years, researchers have been perplexed by the increased risk of almost all causes of death among Alaska Native infants, particularly those residing in Alaska's western and northern regions. High rates of infant death also have been reported in areas of Canada with a similarly high prevalence of the CPT1A variant.14,21 The current analysis suggests that this highly prevalent, stable (as evidenced by Hardy-Weinberg equilibrium18) variant of CPT1A may play a role. This variant may interact with child neglect, parental substance use, poor parental education, or other factors that decrease the likelihood of noticing or seeking care for acute illnesses. Confirmation of these possibilities will require a larger evaluation, which we are undertaking currently. If our preliminary results are upheld, then the situation will represent a substantial opportunity for reducing the historically high and persistent racial disparities in infant mortality rates found in Alaska and throughout the circumpolar region and a true success for newborn screening, through which the high prevalence of the disorder was originally identified.
This study was supported in part by project H18 MC-00004-11 from the Maternal and Child Health Bureau, Health Resources and Services Administration, Department of Health and Human Services.
- Accepted July 20, 2010.
- Address correspondence to Bradford D. Gessner, MD, Alaska Division of Public Health, PO Box 240249, Suite 354, Anchorage, AK 99524. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
- SIDS =
- sudden infant death syndrome •
- CPT1A =
- carnitine palmitoyltransferase type 1A •
- CI =
- confidence interval
- Adams MM
- Schoellhorn KJ,
- Perham-Hester KA,
- Goldsmith YW
- Holman RC,
- Curns AT,
- Cheek JE,
- et al
- Chace DH,
- DiPerna JC,
- Mitchell BL,
- Sgroi B,
- Hofman LF,
- Naylor EW
- Sinclair G,
- Ma J,
- MacLeod P,
- Arbour L,
- Vallance H
- Rajakumar C,
- Ban MR,
- Cao H,
- Young TK,
- Bjerregaard P,
- Hegele RA
- Gessner BD,
- Gillingham MB,
- Johnson MA,
- et al
- Gessner BD,
- Chimonas MAR,
- Grady SC
- Brown NF,
- Mullur RS,
- Subramanian I,
- et al
- 21.↵Statistics Canada. 2006 Census: Infant Mortality by Province and Territory. Ottawa, Canada: Statistics Canada; 2006
Is The Emergency Department the Medical Home?: Who should provide care for an individual with recent onset cough, abdominal pain, or fever that requires medical attention? One would hope that the primary care physician would be the one to see the patient. Unfortunately, as chronicled in an article in The New York Times (Sack K, September 7, 2010), that is too often not the case. Only 42% of the 354 million visits each year in the US for acute illness are to the primary care physician. An examination of emergency department data from 2001 to 2004 showed that 28% of all acute care visits took place in the emergency department while specialists provided 20% of the care for acute illness. Emergency department physicians, however, provided almost all the care for acute illnesses occurring during weekends and after office hours. Almost half of the patients seeking care for acute illness in the emergency department were uninsured, placing a huge burden on hospitals. Reasons for such heavy use of the emergency department are complex and include poor access to primary care. The implications are enormous. While the health care reform act is designed to improve access to primary care and further patient-centered medical homes, the system is already overburdened. Even more people may be showing up in the emergency department for management of their acute illness. As a result, emergency department physicians may have to hone their primary care skills although hopefully health reform will change this pattern and put the emphasis on increasing the primary care medical home in the years ahead.
Noted by JFL, MD and WVR, MD
- Copyright © 2010 by the American Academy of Pediatrics