04 Feb 2022

Tasked with presenting thousands of unique antigens to T cells, the proteins that form the major histocompatibility complex arise from genes on the human leukocyte antigen (HLA) locus—the most polymorphic address in the entire genome. Carriers of certain polymorphs, it turns out, have a protective edge against neurodegeneration, according to a manuscript posted on medRxiv on December 30. Led by Michael Greicius and Emmanuel Mignot at Stanford University, and Jean-Charles Lambert of the University of Lille in France, scientists paired meta-analyses of massive GWAS from around the world with fine mapping of the HLA locus. Their finding? The HLA subtype DRB1*04 protects against both AD and PD. People lucky enough to express it—comprising 30 percent of the European population and slightly less among people of other ancestries—also had fewer tau tangles in their brains at death, and less phospho-tau in their cerebrospinal fluid during life.

Notably, DRB1*04 was found to specifically cradle an acetylated peptide from tau’s aggregation-prone PHF6 region. This implies that this flavor of MHC might bait T cells to mount an immune response against pathological forms of tau.

“This discovery is extremely exciting, and is in line with the narrative introduced more than two decades ago that adaptive immunity plays an important role in recovery from damage to the CNS, a phenomenon that we termed ‘protective autoimmunity,’” Michal Schwartz of the Weizmann Institute of Science in Rehovot, Israel, wrote to Alzforum (Moalem et al., 1999, full comment below).

Penned by 160 authors, the new study was a large collaboration across the fields of immunology, genetics, and neurodegeneration. Its senior author, Mignot, is known for discovering that the HLA allele DQB1*06:02 raises a person's risk for narcolepsy, exposing this sleep disorder as having an autoimmune cause (Matsuki et al., 1992; Mignot et al., 2001; Mignot, 2014). 

HLA genes encode major histocompatibility complex (MHC) class I and class II proteins, each of which is capable of presenting a repertoire of hundreds of unique peptides on the cell surface. CD8+ T cells recognize peptides presented by MHC I, while CD4+ T cells latch on to peptides complexed with MHC II.

While potentially autoreactive T cells that tightly latch onto self-peptides presented by MHC are eliminated during their infancy in the thymus, the system is not foolproof. In the case of narcolepsy, Mignot and colleagues found that DQB1*06:02 presents to CD4+ T cells a peptide torn from hypocretin, a neurotransmitter crucial for wakefulness. When T cells recognized the peptide, they unleashed an autoimmune response that wiped out hypocretin-producing neurons, triggering narcolepsy. In this study, HLA-DRB1*04 is an MHC class II polymorph that presents acetylated tau to CD4+ cells.

This is not the first time HLA genes have been implicated in risk for age-related neurodegenerative diseases. Previous genome-wide association studies had tied polymorphisms at the HLA locus to risk for AD, PD, and ALS (Hollenbach et al., 2019; Naito et al., 2021; Sep 2021 news; van Rheenan et al., 2021). However, the studies lacked consensus on exactly which HLA allele subtypes were responsible. Given the astounding polymorphism that marks the HLA locus, tying a particular GWAS hit to an exact HLA allele requires tenacious genomic sleuthing, in particular, detailed mapping of the HLA locus with all its 3.6 megabases of DNA.

Even so, Mignot’s group recently specified the HLA allele subtype DRB1*04:04 as being most strongly associated with protection against PD in Europeans (Yu et al., 2021). For the current study, the larger collaboration pulled out all the stops, including samples from people of diverse ancestries and expanding their studies to AD GWAS.

First, a brief review of the HLA locus and, yes, the beast of HLA nomenclature. The HLA locus houses 224 genes involved in immune responses, including those that encode the protein subunits that form the six types of major histocompatibility complex. MHC Class I complexes are encoded by HLA-A, B, and C genes, while MHC Class II complexes are encoded by HLA-DP, DQ, and DR genes. These six loci were first identified as transplantation antigens because they can effect rapid rejection of transplanted organs from a mismatched donor. Every MHC molecule presented on the cell surface consists of a paired α- and γ-subunit. The genetic architecture of each HLA type is different. In the-DR subtype, for example, the α-subunit is encoded by a single DRA gene, while the paired β-subunit can be encoded by any of four genes: HLA-DRB1, HLA-DRB3, HLA-DRB4, or HLA-DRB5. However, people only inherit up to three β-subunit genes, which means not everyone does, or even can, express the same DR subtype. To add even more complexity, DRB genes are highly polymorphic, and alleles differing by one or more nucleotides have been identified throughout the global population for each DRB gene. The alleles occur at different frequencies depending on ancestry, and, like any gene, each person can inherit one maternal and one paternal allele for each HLA-DRB gene they express.

These DRB alleles are catalogued by a series of numerical suffixes. The first designates the HLA-DRB allele group, á la HLA-DRB1*01. Alleles in a group share sequences that distinguish them from other allele groups. Within each allele group are subtypes, which typically differ by single nucleotide polymorphisms. These are catalogued by yet an additional suffix, such as in HLA-DRB1*01:04. As much of the polymorphism occurs in regions of the β-subunit that bind to peptides, each HLA allele can present a unique repertoire of thousands of antigenic peptides that at least partially overlap with the repertoire of peptides presented by other alleles in the group. All this boils down to many different DRB1s in the population, but with each person having at most two. Roughly the same goes for DP and DQ genes.

Might other HLA subtypes besides DRB1*04:04 associate with neurodegeneration? First author Yann Le Guen and colleagues started by meta-analyzing 12 PD GWASs from North and South America, Europe, and East Asia—making for a total of 55,554 PD and PD-proxy cases, and 1,454,443 controls. They found single nucleotide polymorphisms (SNPs) in tight linkage with the HLA-DRB1*04 allele group, confirming their previous European findings in a larger and more diverse population. Next, Mignot combed through 12 published AD GWAS and, to his amazement, noticed that the same SNPs—shared by people who express a subtype of the HLA-DRB1*04 allele group—associate with reduced risk for AD. “That’s when we started calling everyone under the sun,” he told Alzforum. Together with Greicius, Lambert, and numerous others, the investigators meta-analyzed AD GWAS data obtained in North America, Europe, Africa, and East Asia, scouring the HLA region in 121,371 AD and proxy-AD cases and in 410,989 controls. They came across the same protective association. Specifically, the rs601945 SNP, found in the HLA-DRB1*04 allele, protected from both PD and AD.

So far, the findings point to an effect in carriers of the HLA-DRB1*04 allele. Within this allele group, several sequence subtypes exist. Might some subtypes provide more protection than others? They found that HLA-DRB1*04:04 and HLA-DRB1*04:07 conferred the strongest protection against both AD and PD, reducing risk by about 15 percent. Weaker effects were seen for HLA-DRB1*04:01 and HLA-DRB1*04:03, and no effect for HLA-DRB1*04:05 or HLA-DRB1*04:06. HLA-DRB1*04:05 is the predominant DRB1*04 subtype in people of Asian ancestry.

The researchers also found hints of a link between HLA-DRB1*04 and protection from other neurodegenerative diseases. In summary data from amyotrophic lateral sclerosis and Lewy body dementia GWAS, the researchers identified protective SNPs that are commonly co-inherited with HLA-DRB1*04. Mignot said the lead SNPs that signify the HLA-DRB1*04-allele group in the AD and PD GWAS were not included within the ALS or LBD GWAS, hence testing if this HLA subtype is protective in all of these disorders will require further imputation.

How might subtypes within the HLA-DRB1*04 allele group protect against neurodegenerative disease? Focusing on AD, the researchers looked for the rs601945 SNP, as proxy for the HLA-DRB1*04 allele group, among more than 7,000 autopsied cases of AD in the Religious Orders Study and Memory and Aging Project (ROSMAP) and the National Institute on Aging–Alzheimer’s Disease Center cohorts. About 29 percent of the cohort carried a DRB1*04 allele, including 16 percent who carried HLA-DRB1*04:01 subtype and nearly 6 percent who carried HLA-DRB1*04:04 subtype. On average, HLA-DRB1*04 carriers had less neurofibrillary tangle burden than did noncarriers. The allele also associated with later age at onset of AD. Furthermore, among more than 8,000 people of European ancestry in the European Alzheimer’s Database, HLA-DRB1*04 carriers had less total tau and phospho-tau in their cerebrospinal fluid than noncarriers. Notably, HLA-DRB1*04 was not tied to burden of amyloid plaques or to CSF Aβ42, suggesting a specific relationship with tau.

The scientists reasoned that HLA-DRB1*04 might bind tau peptides and present them to CD4+ T cells. Most T cells with a penchant for “self” proteins get weeded out by negative selection during their development in the thymus. However, T cells that happen to recognize post-translationally modified versions of self proteins might squeak through. In fact, multiple autoantigens—i.e., the instigators of autoimmune disease—are post-translationally modified peptides. Tau undergoes myriad modifications, and some of them, for example phosphorylation and acetylation, crop up in pathogenic forms of the protein.

To investigate whether HLA-DRB1*04 might bind one of these, the researchers generated more than 1,000 15mer peptides that encompass all tau isoforms, each peptide overlapping by 11 residues. This library included both unmodified and modified peptides found in people with AD, along with some peptides from α-synuclein and some control peptides (Dec 2020 news). The scientists screened the library for binding to the highly protective HLADRB1*04:04 subtype, the moderately protective HLA-DRB1*04:01 subtype, and the neutral HLA-DRB1*04:05 subtype. Few peptides bound to any of the HLA-DRB1*04 peptides. One of them: a piece of tau’s infamous PHF6 domain, which is integral to the core of tau fibrils.

Strikingly, HLADRB1*04 only bound to this tau peptide when it was acetylated at residue K311. In more detailed studies of this peptide, the scientists found that the strength with which it bound to different HLADRB1*04 subtypes matched the hierarchy of the protective effect seen in GWAS, with the HLADRB1*04:04 latching on most tightly, and the HLADRB1*04:05 and HLADRB1*04:06 subtypes binding weakly. The scientists also identified other peptides within tau, as well as α-synuclein, that bound to HLADRB1*04. However, none were modified, making it less likely that autoreactive CD4+ T cells that recognize them would have survived in the thymus.

Present to Protect? In this model, tangle-bearing neurons secrete tau fragments, which are taken up by nearby microglia. The microglia present acetylated tau PHF6 (red) on HLA-DR4 MHC II complexes to CD4+ T cells, which subsequently rally anti-tau antibody responses and clearance of pathological forms of tau. [Courtesy of Le Guen et al., medRxiv, 2021.]

Li Gan of Weill Cornell Medical College, New York, was struck by the selectivity of DRB1*04:04 for the acetylated over the non-acetylated form of PHF6 peptide.

She was impressed at how many modified-tau peptides the researchers screened, but said the screen was limited to tau peptides that had been reported in insoluble forms of tau. Gan and others previously identified acetylated forms of tau within soluble fractions, which were also found to seed aggregation (Sep 2015 news; Apr 2016 news; Trzeciakiewicz et al., 2020). Gan noted that soluble tau fragments are most likely to be processed and presented by MHC Class II molecules. Therefore, further studies could test modified tau forms found in soluble fractions for binding to DRB1*04:04, and also hunt for antibodies specific for acetylated forms of tau in people expressing this and other HLA allele subtypes.

The findings are intriguing in light of previous reports that tau-specific T cells roam the CNS, noted Iryna Prots of University Hospital Erlangen, Germany. “Considering the data of this manuscript and the fact that pronounced post-translational modifications might increase the immunogenicity of the protein, the auto-reactivity of T cells to tau may represent a protective mechanism,” she wrote. “This, in turn, points toward a beneficial autoimmune response in contrast to a classical view on autoimmune reaction as a pathological condition,” she added. Prots noted that the study jibes with model of “protective autoimmunity,” previously proposed by Schwartz (Schwartz and Raposo, 2014). 

To Gan’s mind, the data raise an intriguing hypothesis about how a certain HLA subtype might slow the spread of tau. One question the study left unanswered is how the proposed mechanism would protect against a non-tauopathy such as PD. Mignot acknowledged this conundrum, noting that other mechanisms could be at play in that disease. However, tau becomes post-translationally modified in response to many types of neuronal stress, including a-synuclein pathology. Perhaps eliminating modified forms of tau that influence the formation of other aggregates might stave off progression of non-tauopathies, Mignot proposed. “Tau seems to be the most unifying factor across neurodegenerative diseases,” he said.

If the proposed mechanism underlying HLADRB1*04 protective effect turns out to be true, then peptide vaccines targeting the acetylated PHF6-tau epitope could theoretically slow progression of AD in HLADRB1*04 carriers, Mignot said.—Jessica Shugart

Comments

1. • Michal Schwartz
     Weizmann Institute of Science
   • Posted: 04 Feb 2022

   In this study, the authors attribute an important protective role to self-reactive T cells and speculate that the T cell facilitates early clearance of toxic aggregated tau seeds, either by recruiting β cells and generating an antibody response targeting the same epitope recognized by the T cells, or simply by activating T cell clearance of such early aggregates. Although the antigenic specificity and the mechanism proposed by the authors require further study, the overall observations support the notion that neurodegenerative disease could benefit from adaptive immune response, and thus from immunotherapy.

   This discovery is extremely exciting, and is in line with the narrative introduced more than two decades ago that adaptive immunity plays an important role in recovery from damage to the CNS (Moalem et al., 1999), a phenomenon that we termed “Protective Autoimmunity.” At that time, together with Prof. J. Kipnis, back then a Ph.D. student, we demonstrated that the ability to mount a protective immune response to CNS damage in mice is genetically controlled by the MHC-II loci (Kipnis et al., 2001). These observations, together with numerous additional studies over the last two decades, introduced the role of systemic adaptive immunity in coping with Alzheimer’s disease (AD), and have led to the suggestion of immunotherapy for the treatment of neurodegenerative diseases (Schwartz, 2017). 

   References:

   Moalem G, Leibowitz-Amit R, Yoles E, Mor F, Cohen IR, Schwartz M. Autoimmune T cells protect neurons from secondary degeneration after central nervous system axotomy. Nat Med. 1999 Jan;5(1):49-55. PubMed.

   Kipnis J, Yoles E, Schori H, Hauben E, Shaked I, Schwartz M. Neuronal survival after CNS insult is determined by a genetically encoded autoimmune response. J Neurosci. 2001 Jul 1;21(13):4564-71. PubMed.

   Schwartz M. Can immunotherapy treat neurodegeneration?. Science. 2017 Jul 21;357(6348):254-255. PubMed.

   View all comments by Michal Schwartz
2. • Iryna Prots
     University Hospital Erlangen
   • Posted: 04 Feb 2022

   The genetics of the human leukocyte antigen (HLA) system is very complex and fascinating, and a genetic association with the HLA locus is indicative of an involvement of the adaptive immune system. In this elegant and exciting study, the authors re-analyzed existing genome-wide association studies (GWAS) from an impressive number of participants with Parkinson’s and Alzheimer’s diseases and controls, focusing on fine mapping of the HLA locus. By this approach, a protective association of the HLA-DRB1*04 subtype with PD and AD was identified and was biologically strengthened by HLA-DRB1*04 associating with fewer neurofibrillary tangles in postmortem brains and lower tau levels in the cerebrospinal fluid (CSF).

   The exciting result of preferential binding of the acetylated K311 tau PHF6 peptide, which is prone to aggregate, by the protective HLA-DRB1*04 subtypes potentiates the idea that an adaptive immune response against aggregated tau might be beneficial and protective against neurodegenerative diseases.

   The results of this study have several important implications for deepening our understanding of neurodegenerative mechanisms in tauopathies. I would like to highlight two conceptual aspects: On one hand, the results emphasize a crucial involvement of the adaptive immune response in the pathology of human neurodegenerative diseases; and on the other, this study adds to the concept of protective autoimmunity proposed by Michal Schwartz. While initially the idea of protective autoimmunity was that autoimmune T lymphocytes play a beneficial role in repair of the central nervous system (CNS) after injury, this study expands the potential benefits of an autoimmune reaction against CNS self-antigens to protection against neurodegeneration.

   These conceptually novel findings might revolutionize our knowledge and the development of therapeutics in the field of neurodegenerative disorders. As proposed by the authors, possibilities include either immunizing DRB1*04-carrying individuals with the acetyl-K311 PHF6 epitope, or targeting the acetyl-K311 PHF6 tau epitope with chimeric antigen receptor (CAR) T cells.

   The presence of tau-specific T cells in healthy individuals was reported previously, and was quite surprising. Considering the data in this manuscript, and the fact that pronounced post-translational modifications might increase tau's immunogenicity, the auto-reactivity of T cells to tau may represent a protective mechanism aiming to control tau aggregation and the appearance of neurofibrillary tangles, thereby preventing neurodegeneration onset and/or progression.

   This, in turn, points toward a beneficial autoimmune response, in contrast to the classical view of autoimmune reaction as pathological. In order to remain beneficial, an autoimmune response needs tight control, which might be provided by regulatory cytokines and T cells. In this respect, it would be very interesting to determine which effector T cell populations are activated by the DRB1*04-acetylated-K311 PHF6 peptide presentation, and what is the differentiation capacity of acetyl-K311 PHF6-reacting T cells. In line with this, a comparison of effector T cell profiles in the peripheral blood of AD patients with and without DRB1*04 subtypes could shed light on the balance of pro- and anti-inflammatory (anti-tau) responses and their correlation to the HLA class II subtypes. An open question is, do HLA-DRB1*04 individuals have a less severe or more slowly progressing AD?

   Interestingly, despite the protective association of HLA-DRB1*04 with PD and Lewy pathology reported in this study, no preferential binding of DRB1*04 subtypes to PTM-positive α-synuclein peptides was observed. Does this suggest a central role of tau as an auto-epitope(s) for protective autoimmune response that hamper proteinopathy in different neurodegenerative diseases?

   View all comments by Iryna Prots

Make a Comment

To make a comment you must login or register.

References

News Citations

1. From a Million Samples, GWAS Squeezes Out Seven New Alzheimer's Spots 16 Sep 2021
2. Mounting Modifications Move Tau Toward Aggregation in Alzheimer’s Brain 11 Dec 2020
3. New Type of Toxic Tau? Acetylated Form Correlates With Memory Defects 25 Sep 2015
4. Acetylated Tau Mucks Up Memories 1 Apr 2016

Paper Citations

1. Moalem G, Leibowitz-Amit R, Yoles E, Mor F, Cohen IR, Schwartz M. Autoimmune T cells protect neurons from secondary degeneration after central nervous system axotomy. Nat Med. 1999 Jan;5(1):49-55. PubMed.
2. Matsuki K, Grumet FC, Lin X, Gelb M, Guilleminault C, Dement WC, Mignot E. DQ (rather than DR) gene marks susceptibility to narcolepsy. Lancet. 1992 Apr 25;339(8800):1052. PubMed.
3. Mignot E, Lin L, Rogers W, Honda Y, Qiu X, Lin X, Okun M, Hohjoh H, Miki T, Hsu S, Leffell M, Grumet F, Fernandez-Vina M, Honda M, Risch N. Complex HLA-DR and -DQ interactions confer risk of narcolepsy-cataplexy in three ethnic groups. Am J Hum Genet. 2001 Mar;68(3):686-99. Epub 2001 Feb 13 PubMed.
4. Mignot EJ. History of narcolepsy at Stanford University. Immunol Res. 2014 May;58(2-3):315-39. PubMed.
5. Hollenbach JA, Norman PJ, Creary LE, Damotte V, Montero-Martin G, Caillier S, Anderson KM, Misra MK, Nemat-Gorgani N, Osoegawa K, Santaniello A, Renschen A, Marin WM, Dandekar R, Parham P, Tanner CM, Hauser SL, Fernandez-Viña M, Oksenberg JR. A specific amino acid motif of HLA-DRB1 mediates risk and interacts with smoking history in Parkinson's disease. Proc Natl Acad Sci U S A. 2019 Apr 9;116(15):7419-7424. Epub 2019 Mar 25 PubMed.
6. Naito T, Satake W, Ogawa K, Suzuki K, Hirata J, Foo JN, Tan EK, Toda T, Okada Y. Trans-Ethnic Fine-Mapping of the Major Histocompatibility Complex Region Linked to Parkinson's Disease. Mov Disord. 2021 Aug;36(8):1805-1814. Epub 2021 May 11 PubMed.
7. van Rheenen W, van der Spek RA, Bakker MK, van Vugt JJ, Hop PJ, Zwamborn RA, de Klein N, Westra HJ, Bakker OB, Deelen P, Shireby G, Hannon E, Moisse M, Baird D, Restuadi R, Dolzhenko E, Dekker AM, Gawor K, Westeneng HJ, Tazelaar GH, van Eijk KR, Kooyman M, Byrne RP, Doherty M, Heverin M, Al Khleifat A, Iacoangeli A, Shatunov A, Ticozzi N, Cooper-Knock J, Smith BN, Gromicho M, Chandran S, Pal S, Morrison KE, Shaw PJ, Hardy J, Orrell RW, Sendtner M, Meyer T, Başak N, van der Kooi AJ, Ratti A, Fogh I, Gellera C, Lauria G, Corti S, Cereda C, Sproviero D, D'Alfonso S, Sorarù G, Siciliano G, Filosto M, Padovani A, Chiò A, Calvo A, Moglia C, Brunetti M, Canosa A, Grassano M, Beghi E, Pupillo E, Logroscino G, Nefussy B, Osmanovic A, Nordin A, Lerner Y, Zabari M, Gotkine M, Baloh RH, Bell S, Vourc'h P, Corcia P, Couratier P, Millecamps S, Meininger V, Salachas F, Mora Pardina JS, Assialioui A, Rojas-García R, Dion PA, Ross JP, Ludolph AC, Weishaupt JH, Brenner D, Freischmidt A, Bensimon G, Brice A, Durr A, Payan CA, Saker-Delye S, Wood NW, Topp S, Rademakers R, Tittmann L, Lieb W, Franke A, Ripke S, Braun A, Kraft J, Whiteman DC, Olsen CM, Uitterlinden AG, Hofman A, Rietschel M, Cichon S, Nöthen MM, Amouyel P, SLALOM Consortium, PARALS Consortium, SLAGEN Consortium, SLAP Consortium, Traynor BJ, Singleton AB, Mitne Neto M, Cauchi RJ, Ophoff RA, Wiedau-Pazos M, Lomen-Hoerth C, van Deerlin VM, Grosskreutz J, Roediger A, Gaur N, Jörk A, Barthel T, Theele E, Ilse B, Stubendorff B, Witte OW, Steinbach R, Hübner CA, Graff C, Brylev L, Fominykh V, Demeshonok V, Ataulina A, Rogelj B, Koritnik B, Zidar J, Ravnik-Glavač M, Glavač D, Stević Z, Drory V, Povedano M, Blair IP, Kiernan MC, Benyamin B, Henderson RD, Furlong S, Mathers S, McCombe PA, Needham M, Ngo ST, Nicholson GA, Pamphlett R, Rowe DB, Steyn FJ, Williams KL, Mather KA, Sachdev PS, Henders AK, Wallace L, de Carvalho M, Pinto S, Petri S, Weber M, Rouleau GA, Silani V, Curtis CJ, Breen G, Glass JD, Brown RH Jr, Landers JE, Shaw CE, Andersen PM, Groen EJ, van Es MA, Pasterkamp RJ, Fan D, Garton FC, McRae AF, Davey Smith G, Gaunt TR, Eberle MA, Mill J, McLaughlin RL, Hardiman O, Kenna KP, Wray NR, Tsai E, Runz H, Franke L, Al-Chalabi A, Van Damme P, van den Berg LH, Veldink JH. Common and rare variant association analyses in amyotrophic lateral sclerosis identify 15 risk loci with distinct genetic architectures and neuron-specific biology. Nat Genet. 2021 Dec;53(12):1636-1648. Epub 2021 Dec 6 PubMed. Correction.
8. Yu E, Ambati A, Andersen MS, Krohn L, Estiar MA, Saini P, Senkevich K, Sosero YL, Sreelatha AA, Ruskey JA, Asayesh F, Spiegelman D, Toft M, Viken MK, Sharma M, Blauwendraat C, Pihlstrøm L, Mignot E, Gan-Or Z. Fine mapping of the HLA locus in Parkinson's disease in Europeans. NPJ Parkinsons Dis. 2021 Sep 21;7(1):84. PubMed.
9. Trzeciakiewicz H, Ajit D, Tseng JH, Chen Y, Ajit A, Tabassum Z, Lobrovich R, Peterson C, Riddick NV, Itano MS, Tripathy A, Moy SS, Lee VM, Trojanowski JQ, Irwin DJ, Cohen TJ. An HDAC6-dependent surveillance mechanism suppresses tau-mediated neurodegeneration and cognitive decline. Nat Commun. 2020 Nov 2;11(1):5522. PubMed.
10. Schwartz M, Raposo C. Protective Autoimmunity: A Unifying Model for the Immune Network Involved in CNS Repair. Neuroscientist. 2014 Jan 6;20(4):343-358. PubMed.

Further Reading

No Available Further Reading