Fig. 6.1
Basic approaches to genetic studies of complex diseases
6.2 Basic Concept of Genetics
6.2.1 Structure and Function of Genes and Chromosomes
The gene is the basic unit of heredity of living organisms. It stores the instructions for diverse protein molecules that control development, survival, and reproduction. The whole human genome encodes 20,000–30,000 genes. Deoxyribonucleic acid (DNA) is the molecular component of a gene. The molecular unit of DNA is the nucleotide, which has three basic components: a pentose sugar, a phosphate group, and a nitrogenous base. There are four types of nitrogenous bases: cytosine, thymine, adenine, and guanine. They commonly are represented by their first letters: C, T, A, and G. Nucleotides attach to one another in a certain order to form a polynucleotide chain. Two complementary polynucleotide chains, G pairing with C and T pairing with G, held together by weak thermodynamic forces, form a DNA molecule. Different sequences of nucleotides represent different proteins or different regulatory functions. To encode all the information of the human body, each cell contains approximately three billion nucleotide pairs. To package all this DNA into a tiny cell nucleus, DNA coils around histones in an organized manner and loops into a helical line, thereby forming chromosomes (Fig. 6.2).
Fig. 6.2
Structures of DNA, genes, and chromosomes
For diploid organisms, chromosomes exist in pairs. One member of each pair originates from the father and the other originates from the mother. Each human somatic cell contains 23 pairs of chromosomes, including 22 homologous pairs of autosomes and one pair of sex chromosomes. In normal males, the sex chromosomes are a Y chromosome inherited from the father and an X chromosome inherited from the mother. Two X chromosomes are found in normal females, each inherited from one parent. Two homologues have similar sequences; only a small fraction of positions have sequencing variations that can be used to distinguish the chromosomes. The variants on the same position of paired chromosomes are defined as alleles. The two alleles at a certain position may be the same (homozygous) or different (heterozygous).
The functions of genes may be classified into two general categories: protein-coding and nonprotein-coding genes. The coding sequences of protein-coding genes (exons) are separated from each other by noncoding intervening sequences (introns) (Fig. 6.3). They encode proteins through two major steps: transcription and translation. First, the DNA is transcribed pre-messenger ribonucleic acid (pre-mRNA). Introns then are spliced out, and exons join to form mature messenger RNAs (mRNAs). Second, mRNAs are translated to protein. Every three nucleotides (codons) in mRNA represent an amino acid. Nonprotein-coding genes are transcribed into noncoding RNAs, which are not translated to protein, but form secondary structures to mediate gene expression or mRNA degradation (microRNAs).
Fig. 6.3
Gene structure, transcript, and translation
6.2.2 Genetic Variations and Genetic Markers
Any two human genomes differ at millions of different positions. There are small variations in the individual nucleotides of the genomes, as well as many larger variations, such as microsatellites, insertions, deletions, and copy number variations. Any of these may cause alterations of protein structure or gene expression profile, altering the risk of disease.
One important type of genetic variant is the single-nucleotide polymorphism (SNP). SNPs are a 1 bp substitution of DNA sequences that may be found in any position of the genome. In the human genome, on average, an SNP occurs every 300 nucleotides. Approximately 90 % of the genetic variants in the genome are SNPs, which means there are roughly ten million SNPs in the human genome.
A second major type of variation is the microsatellite, also called a variable number of tandem repeats (VNTR). This term refers to a short nucleotide sequence that occurs as a repeating sequence. The number of repeats differs among different chromosomes in different people, enabling VNTR to be a marker for personal identification.
Another type of genetic variant is the insertion or deletion (indel) of one or more bases in DNA sequences. For protein-coding genes, these changes may result in extra or missing amino acids in a protein if the inserted or deleted sequence is 3 bp long or in a complete change of the protein sequence beyond the indel if the inserted or deleted sequence is not a multiple of 3 (the so-called frameshift). Changes in amino acid sequences may have profound consequences, and many serious genetic diseases are caused by such changes.
On a larger scale, recent studies also focused on many large structural variations in DNA sequences. For example, duplications of large chromosome segments containing one or more genes have been studied widely. The number of duplicates is defined as the copy number, and variation in the number of duplicates among people is termed copy number variation (CNV). Another type of large structure variation is an inversion, a segment of chromosomes reversed end to end. Inversions also have been associated with complex diseases.
A genetic marker is a variable DNA sequence at a known location in the genome. All genetic variants may be used as genetic markers once their locations are confirmed. One of the most important objectives of the HapMap Project is to identify the positions of genetic variants [2]. Genetic markers may be used to study the relationship between an inherited disease and its genetic causes (see details later).
6.2.3 Mutations and Polymorphisms
A mutation is a change in DNA sequence caused by unrepaired damage to DNA or replication errors. Mutations lead to new genetic variants, and their effects range from fatal to mildly detrimental or even beneficial. If a mutation is not fatal, the individual carrying it can reproduce, thereby allowing the mutation to be passed to the next generations and to increase in frequency. For a seriously detrimental mutation, the individual is less likely to survive and reproduce, so the mutation very likely would become extinct over a few generations. Therefore, such deleterious mutations usually are rare in the general population. Genetic variants with a minor allele frequency (MAF) of less than 1 % are classified as rare variants. If the MAF is greater than 1 %, the variant is termed a polymorphism. Among polymorphisms, variants with an MAF greater than 5 % are called common variants; variants with an MAF from 1 % to 5 % are called low-frequency variants. In genetic studies, different methods and technologies are used to assay genetic variants with different frequencies (see later).
6.2.4 Mendelian Genetics and Complex Disease
A genetic disease is a disorder caused by an abnormality in an individual’s DNA. Abnormalities range from a small mutation in a single gene to the addition or subtraction of a subset of chromosomes or even an entire chromosome. There are two major categories of genetic diseases: Mendelian and complex diseases. With regard to Mendelian disorders, mutations in a single gene are sufficient to cause disease. Mendelian disorders are relatively rare and often are first recognized clinically by their predictable patterns of inheritance in families. Common Mendelian modes of inheritance include dominant inheritance, recessive inheritance, and X-linked and Y-linked dominant or recessive modes. More than 4,000 human diseases are caused by single-gene defects. Several skeletal abnormalities follow Mendelian inheritance; classic examples include osteogenesis imperfecta (OI) and spondyloepiphyseal dysplasia.
Mendelian disorders account for only a small proportion of the total burden of human genetic diseases. A much larger component is composed of congenital malformations and common adult diseases. These diseases have significant genetic components and also are influenced by multiple environmental factors. They commonly are called complex diseases or polygenic diseases. Complex disorders are caused by variant forms of genes and environmental factors; they may act independently or modify the effects of one another through gene–gene and/or gene–environment interactions.
LDD is an example of a complex disease. In earlier studies, age, gender, occupation, cigarette smoking, height, and weight were found to be associated with LDD [3]. Later studies suggested a large degree of genetic influences [4]. Recently, many susceptible genes were found to be associated with LDD [5].
6.2.5 Recombination, Linkage, and Linkage Disequilibrium
With regard to the relationship between two genetic variants in a population, some alleles located near each other on the same chromosome are transmitted together, rather than independently during reproduction. Such co-segregated behavior is termed linkage. However, not all genes on the same chromosome are linked, because recombination occurs during meiosis. When homologous chromosomes are paired, crossover occurs between non-sister chromosomes that are part of the chromosome exchanged (Fig. 6.4). As a result of exchange, new combinations of alleles can be formed on a chromosome. The probability of crossover events varies among the different regions of a chromosome. The nearer two positions are, the less chance recombination will occur. The recombination rate is an indication of the distance between two genetic markers, because the further away the markers are, the greater the likelihood of recombination. The distance between two markers may be expressed in centimorgans; 1 cM corresponds to 1 % of recombination and to about one billion bases.
Fig. 6.4
Concepts of recombination
In the human population, crossover positions vary; therefore, the allelic combinations of two markers differ among individuals. Normally, the frequency of non-recombined and recombined alleles is not equal to the randomized frequency. This nonrandom association between markers is called linkage disequilibrium (LD) and is measured by r 2 . The greater the r 2 , the more likely co-segregation will occur between two markers. Thus, the presence of one allele of a marker provides information indicating the allele of the nearer markers by the degree of LD. Recombination and LD are the theoretical basis of the linkage and association studies introduced later.
6.3 Identify the Genetic Cause of Disease
6.3.1 Estimate the Genetic Contribution of LDD
The first step in studies to identify disease genes is to estimate the heritability or the genetic contribution to the disease. This estimation involves observation and statistical analysis of the patterns of phenotypes with various genetic or environmental backgrounds in close kin, such as parent–offspring, siblings, and twins [6]. Familial aggregation and twin studies are two widely used methods for estimating heritability. Familial aggregation estimates the likelihood of a phenotype in close relatives compared with that in controls. On the other hand, studies of twins are useful in estimating the contribution to a phenotype by comparing monozygotic (MZ) pairs (in which all genes are shared) with dizygotic (DZ) pairs (in which half the genes are shared). If the similarity between MZ pairs is greater than that between DZ pairs, the greater part of similarity must be caused by genetics.
Traditionally, disk degeneration was thought to be caused by aging and “wear and tear” from mechanical changes and injuries; however, after familial aggregation and twin studies were conducted, there was a dramatic change in our knowledge of what causes disk degeneration. Several familial aggregation studies conducted in LDD found that young patients with disk herniation had a family history of the disease [7, 8].
The first systematic evaluation of lumbar degenerative changes blinded to twinship was conducted in 1995. Twenty MZ twin pairs from Finland were studied in a pilot investigation of disk degeneration, disk height narrowing, and disk bulging or herniation detected by MRI. A high degree of similarity (26–72 %) was observed between identical twin pairs [4]. Subsequently, a larger and more comprehensive investigation was conducted in 115 pairs of MZ twins. The result showed that 61 % of the variance in disk degeneration was explained by familial aggregation; beyond that, age and occupations requiring heavy lifting together explained 16 % [9]. A 1999 study of female twins in the United Kingdom and Australia enrolled 86 pairs of MZ twins and 77 pairs of DZ twins. The investigators reported 74 % heritability of LDD after adjusting for age, body weight, body height, smoking, occupation, and exercise [10].
6.3.2 Identify the Specific Genes Involved in the Degeneration of Disks
The high heritability estimates for LDD motivated researchers to identify the specific genes responsible. Accurate mapping of the genetic architecture of LDD provides clues to its etiology and pathogenesis. Moreover, those genes might be targets for drug discovery or markers for diagnosis. For Mendelian and complex diseases, the methods for identifying causal genes are different because of the different genetic architectures of these disorders. In Mendelian diseases, causal genes usually are rare and might be identified by studying affected families. Accurate analysis of the mode of disease inheritance might lead to the direct location of the causal variant on the genome. As for complex diseases, the involvement of multiple genes, as well as interactions between genes and the environment, makes it more difficult to identify the causal genetic factors. Two study designs for mapping disease genes are used commonly, namely, family-based design and population-based design; the statistical methods are linkage and association analysis, respectively. A study approach may be classified further as a candidate gene or genome-wide approach depending on whether prior biologic and etiologic knowledge is applied.
6.3.2.1 Linkage Studies
Linkage analysis determines whether the marker segregates with the disease in families with multiple affected individuals, according to a Mendelian mode of inheritance. If a marker is passed down through generations of a family, it may be used as a surrogate for locating adjacent genes. Based on the characteristics of recombination, a linkage study maps disease-causing genes by tracking how alleles of a genetic marker have combined with different members of a family to identify possible crossover regions, thereby indicating shared regions in cases and controls, respectively. Regions shared in cases but not in controls are the candidate positions of disease association genes (Fig. 6.5). Because the crossover event occurs at different positions in different families, a linkage study in multiple families with the same phenotype will narrow down the candidate regions further by overlapping the candidate regions. Microsatellites and SNPs commonly are used as DNA markers in linkage studies.
Fig. 6.5
Linkage analysis
Regarding LDD, a two-stage linkage study in 18 families of southern Chinese descent with early-onset LDD was reported in 2013 [11]. A novel susceptible variant in carbohydrate sulfotransferase 3 (CHST3) was found to be associated with LDD. In the first stage of the study, 400 microsatellites of 89 individuals from 10 families were genotyped. Regions on chromosomes 1, 5, 8, 10, and 20 were identified as candidate regions. In the second stage, 37 individuals from 8 families were added, and another candidate in chromosome 10 was identified. The following association study detected the variant in CHST3 associated with LDD. Additionally, expression of CHST3 mRNA decreased significantly in the intervertebral disk cells of individuals carrying the A allele of SNP rs4148941.
Linkage studies in OI also have been conducted. In one of these studies, an autosomal recessive cause of OI was found in five Turkish families. Linkage mapping demonstrated that all affected individuals shared a 0.83 Mb region on chromosome 17. Further sequencing of this region revealed that the OI phenotype resulted from homozygosity for an in-frame deletion in FKBP10 [12]. A frameshift mutation in transcription factor Osterix, which causes recessive OI, was detected in an Egyptian child [13]. Also found was a missense change, causing autosomal recessive OI [14].
6.3.2.2 Association Studies
Genetic association studies look for a correlation between disease status and genetic variations to identify candidate genes or genome regions that contribute to a specific disease. A genetic association exists if a particular allele is more frequent in the affected group than in the nonaffected group. The greater frequency of this allele in the affected group may indicate that it increases the risk of the disease in question. Statistical analysis usually is performed to determine the significance of the frequency differences. SNPs are the most widely tested genetic markers in association studies, although microsatellite markers, indels, and CNVs also may be used.
A genetic association study usually is conducted in a population-based sample of affected and unaffected individuals (case–control study). After genotyping is performed (Box 6.1), the genotype distributions of each marker are compared between the case and control groups (Fig. 6.6). Statistical measurements of frequency differences then are calculated. A marker that is significantly more frequent in the case versus the control group is identified as a possible disease-causing variant.
Fig. 6.6
Case–control association studies
Association studies may be performed in one of two ways: (i) direct testing of an exposure SNP with known variable phenotypes, such as altered protein structure, and (ii) indirect testing of an SNP, which is a surrogate marker for locating adjacent functional genes that contribute to disease status. The first method requires identification of all variants in the coding and regulatory regions of genes. Although the cost of genotyping continues to decrease, genotyping all ten million SNPs of the whole genome is still expensive. The latter method eliminates the need to catalog all potential susceptibility variants by relying instead on the LD between a genetic marker and susceptible polymorphisms; only a subset of SNPs is genotyped. The genetic constitution of SNPs that are not genotyped may be inferred based on r 2 , which is provided by the HapMap Project [2] (Box 6.1).
Box 6.1. Technologies of SNP Genotyping
The International HapMap Project has enabled researchers to identify most of the common variants across the whole genome [2]. The project also generated allele frequencies and correlations (r 2 ) of the SNPs in different populations. From these data, the LD map was generated as a reference, recording the co-inheritance of the SNPs. The recently launched 1000 Genomes Project aims to provide a more profound characterization of genetic variations in multiple populations [15]. The use of high-throughput sequencing technologies has provided an opportunity to detect not only common variants but also rare variants in different populations. Based on the data generated from these two projects, commercial genotyping microarrays were designed. The genotyping microarray is a solid surface on which artificial microscopic DNA spots of known variants may be attached. The attached DNA spots can capture corresponding sequences marked with an optical signal for identifying genotypes. The Illumina Omni array can detect more than 4.3 million markers, whereas the Affymetrix SNP array can assess 1.8 million markers. Furthermore, the multiplexing Sequenom MassARRAY system and Illumina VeraCode technology have enabled us to genotype a specific region with many SNPs efficiently and cost-effectively. Thanks to these microarrays, a two-step cost-effective approach has been used widely by researchers. The first step is to select a subset of SNPs as a marker for genotyping and to perform an association scan to identify candidate regions. The second step is to genotype a denser set of SNPs within potential regions to look for the exact functional genetic variants.
Candidate Gene Approach
In recent years, the cost of genotyping has dropped dramatically, but the cost of customized genotyping platforms remains high, and limited budgets prevent studies from genotyping a dense set of SNPs across the whole genome. The statistical significance and power of a study are directly affected by the number of individuals tested and by the number of SNPs genotyped. The more SNPs that are genotyped, the more information is obtained and the greater the chance a causal SNP will be identified. To balance cost with study power, investigators select and type the number of SNPs that will maximize the power of the association study. Usually, they do this by choosing a subset of SNPs from candidate genes whose number is commonly determined by the available resources.
The candidate gene approach focuses on selecting genes according to the biological knowledge and etiology of the disease. It takes advantage of our biological understanding of the phenotype, tissues, genes, and proteins that likely are involved in the disease. With regard to disk degeneration, for example, collagen and aggrecan, together with other structural proteins, form the basis of the extracellular matrix, which is an integral part of the disk. These proteins are essential for normal disk function in terms of tensile strength and osmotic pressure. Therefore, the extracellular matrix genes have been prime candidates for genetic study. In addition, gene expression studies are another important way to identify candidate genes. Several tissue homeostasis genes, such as matrix metalloproteinases (MMPs), have elevated expression levels, as well as increased enzymatic activity, during disk degeneration [16]. Moreover, genes identified from other skeletal disorders also may be candidates for LDD. For example, growth differentiation factor 5 (GDF5) is one of the promising candidate genes of osteoarthritis (OA), as it has been reported in multiple populations and has shown high significance [17]. GDF5 was studied as a candidate gene of LDD in northern European women. An SNP (rs143383) was found to be significantly associated with the combination of disk space narrowing and osteophytes [18]. One major drawback of the candidate gene approach, however, is that a priori knowledge of the pathogenesis of the disease is required. If the molecular mechanism is poorly understood, the wrong genes might be selected. Therefore, candidate gene studies are more successful when used as a follow-up to linkage studies [11].
Genome-Wide Association and Meta-Analysis
An advantage of a genome-wide association study (GWAS) is that it requires no prior knowledge of the structure or function of susceptibility genes. A GWAS examines common genetic variants across the whole genome in different individuals to determine whether any variant is associated with disease status. Therefore, this approach makes it possible to identify novel suspected genes. Typically, the power of a GWAS relies on the sample size, definition of the phenotypes, and control of environmental factors. A large sample size, a well-defined phenotype, and proper control of environmental factors might lead to a successful GWAS.
Although GWASs have identified many variants associated with complex diseases, these variants currently explain little about the heritability of most diseases. Normally, the effect sizes of common variants are small, and detection of such small effects requires large sample sizes. Although individual GWASs are underpowered, meta-analyses increase power and reduce false-positive findings. In a meta-analysis, results from several independent studies are contrasted and combined in the hope of identifying the same patterns among study results, sources of disagreement among those results, or other interesting relationships that might come to light in the context of multiple studies.
6.3.2.3 Interpreting the Results
A significant genetic association may be interpreted as (1) a direct association, in which the genotyped SNP is the true causal variant conferring disease susceptibility; (2) an indirect association, in which an SNP in LD with the true causal variant is genotyped; or (3) a false-positive result, in which there is either chance or systematic confounding, such as population stratification. Population stratification is the presence of a systematic difference in allele frequency between subpopulations of the same population, possibly as the result of different ancestry, especially in the context of association studies.
In recent years, the number of genetic studies of LDD has been increasing. It is important to interpret and integrate the results from these studies to improve our overall understanding of LDD, especially with regard to phenotype definitions, statistical significance, and the effects of suspected genes.
Phenotype Definitions
A precise phenotype definition is essential for genetic studies in that the phenotype should be a distinguishable trait and preferably quantifiable. Generally, a trait may be classified as qualitative or quantitative; a qualitative trait can fit into distinct phenotypic categories (case or control), whereas a quantitative trait is measurable as a continuous variable.
With regard to LDD, for example, current assessments of disk degeneration rely on imaging, including radiography and magnetic resonance imaging (MRI). Radiographs may provide information on disk height and osteophyte formation, whereas MRI may show hydration status and disk bulging and herniation, as well as end-plate irregularities. From these images, the presence or absence, or even the severity, of disk degeneration can be defined.
There are several ways to evaluate the degenerative changes in the intervertebral disk. The first is to make a diagnosis based on the presence or absence of disk degeneration, which is an example of a qualitative trait; it is simple and commonly used clinically. However, a disadvantage is that it provides no information about the progressive changes that take place during the degenerative process. Another method is to classify the severity of degeneration based on some well-defined criteria. This method is the one used most widely in genetic studies, and several scoring systems have been developed. For radiographic studies, the Kellgren scale combines the features of osteophytes and joint space narrowing to generate a score ranging from 1, indicating no or very small osteophytes, to 4, representing large osteophytes and pronounced disk space narrowing [19]. For MRI, two scoring systems were developed: the Schneiderman and Pfirrmann scales. Schneiderman’s system focuses on the signal intensity of the nucleus pulposus on MRI, classifying it into four grades, with 0 indicating normal disk with a hyperintense signal (bright disk) and 3 illustrating a hypointense signal with disk space narrowing (black disk) [20]. Pfirrmann’s classification uses MRI images to evaluate the homogeneity of disk structure, signal intensity, and distinction between the nucleus pulposus and annulus fibrosus, as well as disk height. This information is converted into one of five grades, the lowest of which is applied to homogeneous disk structure, hyperintense signal, and normal disk height and the highest to inhomogeneous disk structure, hypointense signal, and loss of distinction between the nucleus pulposus and annulus fibrosus, as well as collapsed disk space [21]. These grading systems provide a semiquantitative evaluation of degenerative status, reflecting the severity of disk degeneration. Interpreting the images from MRI is subjective and thus requires multiple experienced radiologists to perform the grading. The third method for assessing disk degeneration is computational evaluation, which avoids personal errors and saves human resources. Both semiautomated [22] and automated [23] frameworks have been developed for the diagnosis of degenerative disks. The output of computational evaluation is quantitative measurement.
Statistical Significance
The most conspicuous information from a genetic study is whether a genetic variant or gene is associated with a certain disease and may be derived from the test statistics and their corresponding P values. The P value represents the possibility of no association, indicating there is no difference in genotypic distribution between cases and controls. Generally, if P < 0.05, the hypothesis of no association is rejected, meaning there is an association between a genotype and the disease status. In candidate gene studies or GWASs, every genetic variant is tested. If the number of hypotheses being tested increases, the false-positive rate also might increase; therefore, several methods were developed to address this possibility, with the Bonferroni correction being one of the most widely used approaches [24]. The corrected statistical significance level is 1/n times what it would be if only one hypothesis were tested. Thus, the significance threshold of association studies should be 0.05/number of markers being tested.
Effects of Suspect Genes
In epidemiology, relative risk (RR) is the ratio of the probability of a disease occurring in an exposed group to the probability of it occurring in a comparative, nonexposed group (Fig. 6.7a). An RR of N means that the affected group has a risk N times greater than that of the nonaffected group. In genetic association studies, the fundamental unit for reporting effect sizes of suspect genes is the odds ratio (OR). The OR represents the ratio between two proportions: the proportion of individuals in the case group with a specific allele and the proportion of individuals in the control group having the same allele (Fig. 6.7b). An OR >1 demonstrates that individuals carrying the allele are more susceptible to the disease; an OR <1 indicates less susceptibility. The OR is similar to the RR when the disease prevalence is low (in Fig. 6.7, when a and c are small). ORs usually are reported together with 95 % confidence intervals, which show the possible range of a gene’s effect.
Fig. 6.7
Relative risk and odds ratio
6.3.3 Genes Associated with LDD
So far, more than 20 genes have been associated with LDD (Table 6.1), although only a few of them can be replicated, and the results of these studies should be scrutinized closely. In 2008, a Human Genome Epidemiology Network (HuGENet) working group developed a scoring system to provide a reasonable assessment based on three criteria: amount of evidence, replication, and protection from bias. For each criterion, a classification of strong (A), moderate (B), or weak (C) is assigned to the gene study or studies [53]. Eskola et al. [5] used the HuGENet criteria in their systematic review of genetic association studies in LDD. According to their results, most of the associations presented with a weak level of evidence; only five genes showed moderate evidence. None of the studies of disk degeneration genes reached the level of strong credibility. In this chapter, some specific genes are introduced to enhance our understanding of the genetics of LDD.
Table 6.1
Genetic risk factors of lumbar disk degeneration
Gene | Cohort | N | Variant | OR | Phenotype | Reference |
|---|---|---|---|---|---|---|
VDR | Finnish | 85 MZ pairs | TagI and FokI | LDD | Videman et al. [25] | |
Japanese | 205 | TaqI | LDD, LDH | Kawaguchi et al. [26] | ||
Southern Chinese | 804 | TaqI | 2.61 | LDD, LDH | Cheung et al. [27] | |
ACAN | Japanese | 64 | VNTR | Aggrecan protein size | Doege et al. [28] | |
Han Chinese | 132 | VNTR | 1.03–4.5 | LDD | Cong et al. [29] | |
Korean
Related posts: Epidemiology of Lumbar Degenerative Disk Disease
To Fuse or Not to Fuse: That’s the Question
Lessons from a Life: The Journey of Spinal Neurosurgery in the United States
 Minimally Invasive Operation for Lumbar Fusion, Canal Stenosis, Degenerative Scoliosis, and Spondylolisthesis. Is It Possible?
 Degenerative Marrow Changes: Natural History Biomechanics in Relation to Symptoms
 Compensatory Mechanisms Contributing to the Maintenance of Sagittal Balance in Degenerative Diseases of the Lumbar Spine
 Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
Get Clinical Tree app for offline access
|