The Vertebral Artery: A Systematic Review and a Meta-Analysis of the Current Literature

(1) Background. The anatomical variations of the vertebral arteries (VAs) have a significant impact both in neurosurgery and forensic pathology. The purpose of this study was to evaluate the variational anatomy of the vertebral artery. We evaluated anatomical aspects regarding the V1 and V2 segments of the VA: origin, course, tortuosity, hypoplasia, and dominance, and established the prevalence of each variation. (2) Methods. We conducted a systematic search in PubMed and Google Scholar databases, up to December 2022. Sixty-two studies, comprising 32,153 vessels, were included in the current meta-analysis. We used a random-effects model with a DerSimonian-Laird estimator. The confidence intervals were set at 95%. The heterogeneity between studies was assessed using I2. The funnel plot and Egger’s regression test for plot asymmetry were used for the evaluation of publication bias. Statistical significance was considered at p < 0.05. (3) Results. The most common site for the origin of both VAs was the subclavian artery. The aortic arch origin of the left VA had a prevalence of 4.81%. Other origins of the right VAs were noted: aortic arch (0.1%), right common carotid artery (0.1%), and brachiocephalic trunk (0.5%). Ninety-two percent of the VAs entered the transverse foramen (TF) of the C6 vertebra, followed by C5, C7, C4, and least frequently, C3 (0.1%). Roughly one out of four (25.9%) VAs presented a sort of tortuosity, the transversal one representing the most common variant. Hypoplasia occurred in 7.94% of the vessels. Left VA dominance (36.1%) is more common, compared to right VA dominance (25.3%). (4) Conclusions. The anatomy of the VA is highly irregular, and eventual intraoperative complications may be life-threatening. The prevalence of VA origin from the subclavian artery is 94.1%, 92.0% of the VAs entered the TF at C6, 26.6% were tortuous, and 7.94% were hypoplastic.


Introduction
The vertebral arteries (VA), left (LVA) and right (RVA), supply the upper spinal cord, brainstem, cerebellum, and the posterior part of the brain with oxygenated blood, accounting for 28% of its supply [1,2]. Commonly, the VA originates from the first cervical part of the subclavian artery (SA), medial to the anterior scalene muscle, continuing its ascending course through the transverse foramina (TF) of cervical transverse processes, passing through the foramen magnum, to end in the posterior fossa by forming the basilar artery with the opposite VA [3].
The study aimed to assess the morphology of the VA's V1 and V2 segments from a qualitative perspective. There were evaluated the course of the VA, the origin, the vertebral level of entrance in the TF, tortuosity and straightness, planes of tortuosity, hypoplasia, and dominance of this artery.

Materials and Methods
The study was performed according to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines for reporting systematic reviews and meta-analyses [12].

Search Method
Articles were obtained by performing systematic searches on different databases, PubMed and Web of Science, using the following keywords: "Vertebral artery" AND "Anatomy" AND "V1" and "Vertebral artery" AND "Anatomy" AND "V2". The reference list of all the relevant articles was thoroughly scrutinized for other appropriate references to be included in the analysis. Extensive searches were also performed on Google Scholar, Google, and Research Gate. Every potentially relevant article was obtained as a full article, analyzed by two reviewers, and included in the current paper if it fulfilled the inclusion criteria and did not fulfill any of the exclusion criteria. Selected articles were imported into a Paperpile database.

Selection Criteria
The following exclusion criteria were used: (1) articles that did not mention any relevant measurements regarding the first two segments of the VA, (2) results duplicating previously published articles, (3) low-quality articles comprising inadequate methods of measurement and irrelevant results or published in unknown/low impact factor journals, (4) less than 10 subjects, (5) studies conducted on fetuses, (6) inability to obtain the full article, (7) studies that measured the prevalence of tortuosity in patients with Loeys-Dietz syndrome, Marfan syndrome or any other vascular disease, as the frequency of tortuous vessels in these subjects is known to be much higher compared to healthy subjects [13,14], (8) case reports, reviews, meta-analyses, or any other prevalence studies based on published values.

Data Collection and Analysis
Two reviewers separately extracted data from each study in different Excel 365 databases. If any major discrepancy appeared, a third reviewer was co-opted to check the inconsistency and select the appropriate result. The following information was extracted: name of the first author, year, method (CT, autopsy reports, surgical reports, etc.), total number of cases, total number of left and right VAs (LVAs and RVAs), the analyzed segment of the VA (V1 or V2) and its measurements, each in a separate, specific column.

Quality Assessment and Risk of Bias
Newcastle-Ottawa Scale (NOS) for case-control studies was used, with the following adjustments for a prevalence analysis: item 3 from Selection and Exposure was excluded. Each article was noted with a mark ranging from 0 to 7 to assess the quality. Articles with at least 4 points were included in the current paper.

Statistical Analysis
Results were statistically analyzed using Jamovi 2.3 software (Sydney, Australia) for data analysis. Three columns were created: authors, total cases, and the third one which varied, depending on the measured variable. The "Proportions" tab was used to process the variables. The "Total cases" column also varied in accordance with the variable. If the analyzed variable concerned only one side (left or right), then the standardized/true prevalence for that particular side was determined. The following analyses were conducted by determining the crude prevalence for each anatomical variant: C3, C4, C5, C6, and C7 TF entrance, overall tortuosity, VA dominance, and hypoplasia. VA origin, TF entrance corelated to VA origin and plane of tortuosity were assessed as standardized prevalence, as well as the results reported in Table 1, based on study type (autopsy or imaging). Cases that did not mention the exact number of VAs (e.g., 70 specimens) were analyzed carefully. If, after scrutinizing the entire article, the published data indicated a number of arteries double the mentioned one, we considered the first as the real number of VAs (70 LVAs and 70 RVAs) and the latter as the number of patients [15]. The following configuration was used for the statistical processing of the data. The model estimator was DerSimonian-Laird, Raw Proportion was used for effect size model measures, Moderator Type was set on no moderator, and the Confidence Interval (CI) level was established at 95%. Statistical significance was considered at p < 0.05. The funnel plot and Egger's regression test for plot asymmetry were used for the analysis of the publication bias. I 2 was used to test the presence of heterogeneity between studies, using the following thresholds: 0-35%-most likely not important, 36-55%-moderate heterogeneity, 56-85%-most likely substantial heterogeneity, and 86-100%-significant heterogeneity (average values based on [16]). Considering VA course prevalence, Trattnig et al., (1993) [17] report two types of results, measured on two different groups of subjects (140 and 160 subjects, respectively) and with different methods (cadavers and imagistic). The article was, therefore, used twice in the analyses. Regarding hypoplasia, as each author defines a hypoplastic VA differently, we adopted the highest reported value, to include all the reported measurements. Thus, hypoplasia is interpreted as a diameter of 3.5 mm or less. If reported data of hypoplasia or dominance were split into V1 and V2 segments, they were summed up, so the prevalence of hypoplastic/dominant VA was calculated as a whole for the mentioned segments.
Vaiman et al. [18] analyzed a total of 400 VAs, with 23 anomalous VAs. We used 23 as the total number of vessels to determine the prevalences, since the article reported data about the origin and TF entrance only for these anomalous VAs. For each description of the forest plot in the Result section, we reported the number of studies, the total number of vessels included in the analysis, and the number of vessels displaying the anatomical variant. Then, we added the results of the statistical analysis: prevalence, CI, p-value, heterogeneity, and Egger's regression test for plot asymmetry. We also evaluated the standardized prevalence disparity between autopsy and imaging studies and specified whether it was significant or not.

Search Synthesis
During the initial research via databases and other methods, 60 references were gathered for V1 segment and 93 for V2. After removing all articles that displayed at least one exclusion criterion, 62 papers were further scrutinized and included in the current meta-analysis. The search synthesis is systemized in Figure 2. Each paper contained in the present study is detailed in Table 2.

Quality Assessment and Risk of Bias
Each article was assessed with a score between 0 and 7. The score for each study was noted in Table 2. There was no significant bias in any of the mentioned papers.  Figure 3. The heterogeneity was substantial (I 2 = 80.7%). Based on Egger's regression test for plot asymmetry, there was a significant publication bias (p < 0.001). No major difference was found between standardized prevalence determined in autopsy (94.5%) and imaging (93.5%) studies (Table 1).       12,456 LVAs, contained data about the origin of the LVA. The standardized prevalence of the LVA arising from the SA was 94.1% (CI: 93.0-95.2, p < 0.001), as shown in Figure 3. The heterogeneity was substantial (I 2 = 80.7%). Based on Egger's regression test for plot asymmetry, there was a significant publication bias (p < 0.001). No major difference was found between standardized prevalence determined in autopsy (94.5%) and imaging (93.5%) studies (Table 1).  Figure 4. The heterogeneity was substantial (I 2 = 59.41%). Considering Egger's regression test for plot asymmetry, there is a significant publication bias (p < 0.001). We did not identify any considerable difference between the prevalence determined in autopsy (5.47%) and imaging (4.59%) studies (Table 1).  Figure 4. The heterogeneity was substantial (I 2 = 59.41%). Considering Egger's regression test for plot asymmetry, there is a significant publication bias (p < 0.001). We did not identify any considerable difference between the prevalence determined in autopsy (5.47%) and imaging (4.59%) studies (Table 1).  The origin of the RVA from the SA (normal or aberrant) was inspected in twentythree studies. The total amount of investigated RVAs was 8002. The determined prevalence was 99.9% (CI: 99.8-99.9, p < 0.001), as shown in Figure 5. The heterogeneity was not important (I 2 = 0%). Regarding Egger's plot asymmetry regression test, a significant publication bias was found (p < 0.001). No significant prevalence divergence was discovered in autopsy (99.8%) and imaging studies (99.9%) ( Table 1). was found (p < 0.001). No significant prevalence divergence was discovered in autopsy (99.8%) and imaging studies (99.9%) ( Table 1). The origin of the RVA from the SA (normal or aberrant) was inspected in twentythree studies. The total amount of investigated RVAs was 8002. The determined prevalence was 99.9% (CI: 99.8-99.9, p < 0.001), as shown in Figure 5. The heterogeneity was not important (I 2 = 0%). Regarding Egger's plot asymmetry regression test, a significant publication bias was found (p < 0.001). No significant prevalence divergence was discovered in autopsy (99.8%) and imaging studies (99.9%) ( Table 1).

RVA Originating from the AA
In total, 2992 RVAs were examined in four different original studies that reported the presence of an RVA arising from the AA. This type of origin was only found in 5 RVAs. The prevalence was not statistically significant (p = 0.306). The heterogeneity was not important (I 2 = 11.7%) and the publication bias was not statistically significant (p = 0.069), as shown in Table 3.

RVA Originating from the Right Common Carotid Artery
Four articles, analyzing 4368 RVAs, described 8 vessels arising from the right common carotid artery (RCCA). The overall prevalence was 0.126% (CI: 0-0.2, p = 0.019), presented in Table 3. The heterogeneity was not important (I 2 = 0%). Based on Egger's regression test for plot asymmetry, there was not a significant publication bias (p = 0.130).

C3 TF Entrance Prevalence C3 TF Entrance of the LVAs Originating from the AA
Out of forty LVAs that originated from the AA, four of them entered the TF at the level of the third vertebra. The prevalence was 8.11% (CI: −0.2-16.4, p = 0.057), as in Table 3. The heterogeneity was not important (I 2 = 0%). Based on Egger's regression test for plot asymmetry, there was not a significant publication bias (p = 0.280).

C3 TF Entrance
Considering the crude prevalence of the VAs entering the TF at the level of C3, eleven articles, adding up to 7449 VAs, described 15 vessels presenting such an anatomical variant (Figures 6, S1 and S2). The prevalence (0.1%, CI: 0-0.2) was statistically significant (p = 0.008). The heterogeneity was not important (I 2 = 0%). Publication bias measured with Egger's regression test for plot asymmetry was significant (p = 0.009).

C4 TF Entrance Prevalence
C4 TF Entrance of the LVAs Originating from the AA Eight articles described forty-one out of 278 LVAs arising from the AA that entered the TF at the level of the fourth vertebra. The overall prevalence was 14.1% (CI: 10-18.2, p < 0.001), as in Figure 7. The heterogeneity was not important (I 2 = 0%). Publication bias was not statistically significant (p = 0.514).

C4 TF Entrance Prevalence
C4 TF Entrance of the LVAs Originating from the AA Eight articles described forty-one out of 278 LVAs arising from the AA that entered the TF at the level of the fourth vertebra. The overall prevalence was 14.1% (CI: 10-18.2, p < 0.001), as in Figure 7. The heterogeneity was not important (I 2 = 0%). Publication bias was not statistically significant (p = 0.514).
C4 TF Entrance of the LVAs Originating from the SA Four original papers presented nine out of 2736 LVAs arising from the SA, which entered the TF at C4. The prevalence was 0.337% (CI: −0.2-0.9, p = 0.216). The heterogeneity was moderate (I 2 = 39.11%). The publication bias was statistically significant (p = 0.031) ( Table 3).

C4 TF Entrance of the LVAs
In total, 6405 LVAs analyzed in thirteen papers described sixty-three which entered the TF at C4. The overall prevalence was 0.807% (CI: 0.6-1.0, p < 0.001), as in Figures 8, S3 and S4. The heterogeneity was not important (I 2 = 0%). Considering Egger's regression test for plot asymmetry, there is a significant publication bias (p = 0.009).

C4 TF Entrance Prevalence
C4 TF Entrance of the LVAs Originating from the AA Eight articles described forty-one out of 278 LVAs arising from the AA that entered the TF at the level of the fourth vertebra. The overall prevalence was 14.1% (CI: 10-18.2, p < 0.001), as in Figure 7. The heterogeneity was not important (I 2 = 0%). Publication bias was not statistically significant (p = 0.514). C4 TF Entrance of the LVAs Originating from the SA Four original papers presented nine out of 2736 LVAs arising from the SA, which entered the TF at C4. The prevalence was 0.337% (CI: −0.2-0.9, p = 0.216). The heterogeneity was moderate (I 2 = 39.11%). The publication bias was statistically significant (p = 0.031) ( Table 3).

C4 TF Entrance of the LVAs
In total, 6405 LVAs analyzed in thirteen papers described sixty-three which entered the TF at C4. The overall prevalence was 0.807% (CI: 0.6-1.0, p < 0.001), as in Figures 8, S3

C4 TF Entrance of the RVAs
The TF entrance of 7183 RVAs was analyzed in 15 studies. Eighty-seven were seen entering at the level of the C4 vertebra. The prevalence was 1.14% (CI: 0.6-1.6, p < 0.001), shown in Figures 9, S5 and S6. The heterogeneity was substantial (I 2 = 77.99%). Egger's regression test for plot asymmetry showed a significant publication bias (p < 0.001).

C4 TF Entrance of the RVAs
The TF entrance of 7183 RVAs was analyzed in 15 studies. Eighty-seven were seen entering at the level of the C4 vertebra. The prevalence was 1.14% (CI: 0.6-1.6, p < 0.001), shown in Figures 9, S5 and S6. The heterogeneity was substantial (I 2 = 77.99%). Egger's regression test for plot asymmetry showed a significant publication bias (p < 0.001).

C4 TF Entrance of the RVAs
The TF entrance of 7183 RVAs was analyzed in 15 studies. Eighty-seven were seen entering at the level of the C4 vertebra. The prevalence was 1.14% (CI: 0.6-1.6, p < 0.001), shown in Figures 9, S5 and S6. The heterogeneity was substantial (I 2 = 77.99%). Egger's regression test for plot asymmetry showed a significant publication bias (p < 0.001).

C4 TF Entrance
Twenty-five articles reported the entrance of 184 VAs (out of 12,528 VAs) in the TF of the fourth cervical vertebra. The crude prevalence was estimated at 1.25% (CI: 0.9-1.6, p < 0.001), as in Figures 10, S7 and S8. The heterogeneity was moderate (I 2 = 55.23%). Egger's regression

C5 TF Entrance Prevalence
C5 TF Entrance of the LVAs Originating from the AA Ten original papers analyzed 313 LVAs with AA origin. The prevalence of entrance in the TF of the fifth vertebra was 69.2% (CI: 60.8-77.5, p < 0.001), represented in Figure 11. The heterogeneity was moderate (I 2 = 55.69%). The publication bias was not significant (p = 0.177).  Figure 11. The heterogeneity was moderate (I 2 = 55.69%). The publication bias was not significant (p = 0.177).

C5 TF Entrance Prevalence
C5 TF Entrance of the LVAs Originating from the AA Ten original papers analyzed 313 LVAs with AA origin. The prevalence of entrance in the TF of the fifth vertebra was 69.2% (CI: 60.8-77.5, p < 0.001), represented in Figure 11. The heterogeneity was moderate (I 2 = 55.69%). The publication bias was not significant (p = 0.177).

C5 TF Entrance of the LVAs
In total, 7359 LVAs examined in seventeen studies were included in the current evaluation. The prevalence of the TF entrance at C5 was 5.51% (CI: 4.1-6.9, p < 0.001), as in Figures 13, S11 and S12. The heterogeneity was substantial (I 2 = 82.74%). Considering Egger's regression test for plot asymmetry, there was a significant publication bias (p < 0.001).

C5 TF Entrance of the LVAs
In total, 7359 LVAs examined in seventeen studies were included in the current evaluation. The prevalence of the TF entrance at C5 was 5.51% (CI: 4.1-6.9, p < 0.001), as in Figures 13, S11 and S12. The heterogeneity was substantial (I 2 = 82.74%). Considering Egger's regression test for plot asymmetry, there was a significant publication bias (p < 0.001). C5 TF Entrance of the LVAs In total, 7359 LVAs examined in seventeen studies were included in the current evaluation. The prevalence of the TF entrance at C5 was 5.51% (CI: 4.1-6.9, p < 0.001), as in Figures 13, S11 and S12. The heterogeneity was substantial (I 2 = 82.74%). Considering Egger's regression test for plot asymmetry, there was a significant publication bias (p < 0.001).

C5 TF Entrance
In total, 12,664 VAs, examined in thirty articles, were included in the present analysis. The crude prevalence of the VA entering the TF at C5 was 5.87% (CI: 4.4-7.3, p < 0.001), as shown in Figures 15, S15 and S16. The heterogeneity was significant (I 2 = 92.73%). Based on Egger's regression test for plot asymmetry, there was a significant publication bias (p < 0.001). We discovered a notable difference between prevalences determined in autopsy (4.37%) and imaging (8.36%) studies (Table 1).

C5 TF Entrance
In total, 12,664 VAs, examined in thirty articles, were included in the present analysis. The crude prevalence of the VA entering the TF at C5 was 5.87% (CI: 4.4-7.3, p < 0.001), as shown in Figures 15, S15 and S16. The heterogeneity was significant (I 2 = 92.73%). Based on Egger's regression test for plot asymmetry, there was a significant publication bias (p < 0.001). We discovered a notable difference between prevalences determined in autopsy (4.37%) and imaging (8.36%) studies (Table 1).

C5 TF Entrance
In total, 12,664 VAs, examined in thirty articles, were included in the present analysis. The crude prevalence of the VA entering the TF at C5 was 5.87% (CI: 4.4-7.3, p < 0.001), as shown in Figures 15, S15 and S16. The heterogeneity was significant (I 2 = 92.73%). Based on Egger's regression test for plot asymmetry, there was a significant publication bias (p < 0.001). We discovered a notable difference between prevalences determined in autopsy (4.37%) and imaging (8.36%) studies (Table 1).

C6 TF Entrance
Thirty-six papers, including 18,002 VAs, were examined and included in the current analysis. The crude prevalence of the VA entering the TF at the C6 level was 92.0% (CI: 90.5-93.4, p < 0.001), presented in Figure 20. The heterogeneity was significant (I 2 = 91.63%). Based on an Egger's regression test for plot asymmetry, there was a significant publication bias (p = 0.001). A prevalence difference of 5.4% was detected, comparing autopsy (88.2%) and imaging (93.6%) studies (Table 1).

C6 TF Entrance
Thirty-six papers, including 18,002 VAs, were examined and included in the current analysis. The crude prevalence of the VA entering the TF at the C6 level was 92.0% (CI: 90.5-93.4, p < 0.001), presented in Figure 20. The heterogeneity was significant (I 2 = 91.63%). Based on an Egger's regression test for plot asymmetry, there was a significant publication bias (p = 0.001). A prevalence difference of 5.4% was detected, comparing autopsy (88.2%) and imaging (93.6%) studies (Table 1).

C7 TF Entrance of the RVAs
Nine papers, adding up to 4182 RVAs, were included to determine the C7 TF entrance prevalence of the right VAs. The overall value was 0.994% (CI: 0.3-1.7, p = 0.006), as in Figure 23. The heterogeneity was substantial (I 2 = 81.03%). The publication bias, determined using Egger's regression test for plot asymmetry, was significant (p < 0.001).

C7 TF Entrance of the RVAs
Nine papers, adding up to 4182 RVAs, were included to determine the C7 TF entrance prevalence of the right VAs. The overall value was 0.994% (CI: 0.3-1.7, p = 0.006), as in Figure 23. The heterogeneity was substantial (I 2 = 81.03%). The publication bias, determined using Egger's regression test for plot asymmetry, was significant (p < 0.001). C7 TF Entrance of the RVAs Nine papers, adding up to 4182 RVAs, were included to determine the C7 TF entrance prevalence of the right VAs. The overall value was 0.994% (CI: 0.3-1.7, p = 0.006), as in Figure 23. The heterogeneity was substantial (I 2 = 81.03%). The publication bias, determined using Egger's regression test for plot asymmetry, was significant (p < 0.001).

Tortuosity
Overall Tortuosity Fifteen original papers, summing 3149 VAs, were examined to determine the crude tortuosity prevalence of the V1 and V2 segments. Two articles reported different values for "significant" or "high" tortuosity [60,69]. These values were added up to the tortuosity value and analyzed altogether. The prevalence was estimated at 26.6% (CI: 21.4-31.7, p < 0.001), as represented in Figure 25. The heterogeneity was significant (I 2 = 98.8%). The publication bias, based on Egger's regression test for plot asymmetry, was significant (p < 0.001). A prevalence difference of 5.2% was found while comparing autopsy (31.0%) and imaging (25.8%) articles (Table 1).  Fifteen original papers, summing 3149 VAs, were examined to determine the crude tortuosity prevalence of the V1 and V2 segments. Two articles reported different values for "significant" or "high" tortuosity [60,69]. These values were added up to the tortuosity value and analyzed altogether. The prevalence was estimated at 26.6% (CI: 21.4-31.7, p < 0.001), as represented in Figure 25. The heterogeneity was significant (I 2 = 98.8%). The publication bias, based on Egger's regression test for plot asymmetry, was significant (p < 0.001). A prevalence difference of 5.2% was found while comparing autopsy (31.0%) and imaging (25.8%) articles (Table 1).

Overall Tortuosity
Fifteen original papers, summing 3149 VAs, were examined to determine the crude tortuosity prevalence of the V1 and V2 segments. Two articles reported different values for "significant" or "high" tortuosity [60,69]. These values were added up to the tortuosity value and analyzed altogether. The prevalence was estimated at 26.6% (CI: 21.4-31.7, p < 0.001), as represented in Figure 25. The heterogeneity was significant (I 2 = 98.8%). The publication bias, based on Egger's regression test for plot asymmetry, was significant (p < 0.001). A prevalence difference of 5.2% was found while comparing autopsy (31.0%) and imaging (25.8%) articles (Table 1). Figure 25. Forest plot-Tortuosity prevalence. Figure 25. Forest plot-Tortuosity prevalence.

Coronal Tortuosity Prevalence
Three papers, adding up to 380 tortuous VAs, were included in the analysis of coronal tortuosity prevalence. The determined value is 16.6% (CI: 8.4-24.8, p < 0.001), as in Table  The heterogeneity was substantial (I 2 = 78.87%). The publication bias was significant (p < 0.001).

VA Hypoplasia Prevalence
Fourteen articles were included, adding up to 9570 VAs. The crude prevalence of such a condition was 7.94% (CI: 5.6-10.3, p < 0.001), shown in Figure 28. The heterogeneity was significant (I 2 = 94.42%). The publication bias was not significant (p = 0.425). No major difference was detected between autopsy and imaging articles (Table 1).

VA Hypoplasia Prevalence
Fourteen articles were included, adding up to 9570 VAs. The crude prevalence of such a condition was 7.94% (CI: 5.6-10.3, p < 0.001), shown in Figure 28. The heterogeneity was significant (I 2 = 94.42%). The publication bias was not significant (p = 0.425). No major difference was detected between autopsy and imaging articles (Table 1).

VA Hypoplasia Prevalence
Fourteen articles were included, adding up to 9570 VAs. The crude prevalence of such a condition was 7.94% (CI: 5.6-10.3, p < 0.001), shown in Figure 28. The heterogeneity was significant (I 2 = 94.42%). The publication bias was not significant (p = 0.425). No major difference was detected between autopsy and imaging articles (Table 1).

Anatomical Incidence
Considering the VA topography and morphology, the results are in accordance with the information stated in different anatomical textbooks [76,77]. The most prevalent site of origin for both LVA and RVA is the SA with a frequency of 94.1%, respectively 99.9%. Regarding the entrance in a certain TF, most VAs (92%) enter at the level of the sixth cervical vertebra, followed by C5, C7, C4, and, less frequently, C3. AA-originated LVAs present the highest frequency of entrance into the fifth cervical vertebra TF (69.2%), then C4, C6, C3, and C7. This result contradicts Morris [78] who stated that AA-originated LVAs commonly enter the TF of

Anatomical Incidence
Considering the VA topography and morphology, the results are in accordance with the information stated in different anatomical textbooks [76,77]. The most prevalent site of origin for both LVA and RVA is the SA with a frequency of 94.1%, respectively 99.9%. Regarding the entrance in a certain TF, most VAs (92%) enter at the level of the sixth cervical vertebra, followed by C5, C7, C4, and, less frequently, C3. AA-originated LVAs present the highest frequency of entrance into the fifth cervical vertebra TF (69.2%), then C4, C6, C3, and C7. This result contradicts Morris [78] who stated that AA-originated LVAs commonly enter the TF of the fourth cervical vertebra. Roughly one out of four VAs presented a form of tortuosity, the transverse tortuosity having the highest prevalence. From the current meta-analysis, it also resulted in a higher frequency of LVA dominance compared to RVA, which suggests that a greater volume of blood is conducted through the LVA to the brain. Of significant clinical importance is the fact that LVA damage may lead to deterioration of the brain's blood supply [79,80]. As regards hypoplastic VAs, the prevalence is relatively low; roughly 8% of the vessels were reported with a diameter lower than 3.5 mm. VA origin has been widely evaluated and it is acknowledged that the most frequent origin on each side is from the SA. Arteries originating from any other site are considered aberrant [81,82]. LVAs most frequent anatomic variant is the direct aortic origin, with a prevalence of 2.4-5.8% [83]. In the current study, we determined a prevalence of 4.81% of the AA-originated LVA, which conforms with the range 2.4-4.8% [83] reported by Adachi et al. [19], Yamaki et al. [73], Uchino et al. [69], etc. Other origins have been reported: left external carotid artery, thyrocervical trunk, carotid bulb, left common carotid artery, but the frequency is extremely low (0.1-0.2%) [82] and there is an insufficient amount of prevalence studies that reported these variants. Concerning RVA's abnormal origin, the brachiocephalic trunk was the most frequent with a prevalence of 0.539%, followed by RCCA and AA origin. As these variants are remarkably uncommon, they have been mostly communicated as case reports [84][85][86][87][88][89].
As documented in Bergman's Comprehensive Encyclopedia of Human Anatomic Variation, VAs could originate from any carotid artery, common, internal, or external [90]. However, these anatomical possibilities correspond to persistent proatlantal arteries which are rarely found [91] and could appear as either V3 segments with carotid origin, with normoplastic or hypoplastic V1-V2 VAs supplied from the SAs [92], or as proatlantal arteries with absent VAs [93].
The V2 segment of the VA is commonly situated between C6 and C2. An anomalous entrance is frequently correlated with an abnormal origin [69]; the most prevalent variant is the AA-originated LVA entering the fifth cervical TF in 78.6% (Figure 29) [72], compared to 69.2% stated in the current study. Li et al. [50] confirmed this variant, as the LVA with an anomalous origin entered the TF at C5 and presented a sinuous course. Tardieu et al. [94] also found that AA-originating LVAs were more likely to enter the TF more cranially than C6 and to adopt a more medial course over the cervical vertebral bodies.
amount of prevalence studies that reported these variants. Concerning RVA's abnormal origin, the brachiocephalic trunk was the most frequent with a prevalence of 0.539%, followed by RCCA and AA origin. As these variants are remarkably uncommon, they have been mostly communicated as case reports [84][85][86][87][88][89].
As documented in Bergman's Comprehensive Encyclopedia of Human Anatomic Variation, VAs could originate from any carotid artery, common, internal, or external [90]. However, these anatomical possibilities correspond to persistent proatlantal arteries which are rarely found [91] and could appear as either V3 segments with carotid origin, with normoplastic or hypoplastic V1-V2 VAs supplied from the SAs [92], or as proatlantal arteries with absent VAs [93].
The V2 segment of the VA is commonly situated between C6 and C2. An anomalous entrance is frequently correlated with an abnormal origin [69]; the most prevalent variant is the AA-originated LVA entering the fifth cervical TF in 78.6% (Figure 29) [72], compared to 69.2% stated in the current study. Li et al. [50] confirmed this variant, as the LVA with an anomalous origin entered the TF at C5 and presented a sinuous course. Tardieu et al. [94] also found that AA-originating LVAs were more likely to enter the TF more cranially than C6 and to adopt a more medial course over the cervical vertebral bodies.  Gantwerker et al. [95] correlated VA's diameter with the size of the TF; thus, a small TF may presume a hypoplastic or even an aplastic vessel ( Figure 30). Hypoplastic VAs have been widely analyzed for their clinical appliance, being a predisposing factor for posterior circulation ischemia [96]. George and Laurian [97] reported a frequency of 5.7% for LVA hypoplasia and 8.8% for hypoplastic RVAs, without mentioning the hypoplastic segment (V1-V4) and including also the atretic arteries. Gantwerker et al. [95]correlated VA's diameter with the size of the TF; thus, a small TF may presume a hypoplastic or even an aplastic vessel ( Figure 30). Hypoplastic VAs have been widely analyzed for their clinical appliance, being a predisposing factor for posterior circulation ischemia [96]. George and Laurian [97]reported a frequency of 5.7% for LVA hypoplasia and 8.8% for hypoplastic RVAs, without mentioning the hypoplastic segment (V1-V4) and including also the atretic arteries. VA loops occur most frequently in the V2 segment (90.5%), followed by V1 (7.6%) and V3 (1.9%) (Figure 31), with the highest occurrence in females, aged between fifty and seventy years old [98]. VA loops occur most frequently in the V2 segment (90.5%), followed by V1 (7.6%) and V3 (1.9%) (Figure 31), with the highest occurrence in females, aged between fifty and seventy years old [98]. VA loops occur most frequently in the V2 segment (90.5%), followed by V1 (7.6% and V3 (1.9%) (Figure 31), with the highest occurrence in females, aged between fifty and seventy years old [98]. It worth discussing here that attributing other meanings to generally accepted labels, such as V1, V2, or V3, would lead to misunderstandings and erroneous assumption of the reported data. Kiresi et al. [99] and Cacciola et al. [100] described the anatomical area between C3 and C2 TF as V1, C2 TF to C1 TF as V2, and C1 to dural entry point as V3. Fisher et al. [101] describe the V2 segment as the VA part running between C2-C3 and dura mater. This assignment was also used by Lang et al. [102]. Considering the convention of term usage, V1-V4 labels describe the four segments of the VA, from the SA to the origin of the basilar artery [7,8,[103][104][105], as detailed in the Introduction. We strongly advise authors to avoid term reassignments.

Embryology
During embryogenesis, from the dorsal aspect of the aorta arise seven intersegmental branches, referred to as C1-CUsually, the first part of the VA develops from the dorsal branch of the seventh cervical intersegmental artery. Longitudinal communications of the postcostal anastomoses form the second part of the VA [106,107].
The AA-originating LVA is considered to result from a persisting sixth cervical intersegmental artery, which fails to disappear. The blood flows directly from the AA to the C6 intersegmental artery, creating the AA origin of the LVA [108,109]. Increased resorp-tion of embryonic tissue of the LSA between the AA and LVA origin may also lead to an anomalous origin [110].
It should be pointed out the significance of two persistent arteries, proatlantal and hypoglossal artery, which along with the trigeminal and otic arteries represent the deficient involution of embryonic vascular channels [111]. Hypoplasia of the VA, uni-or bilateral, is correlated with a prevalence of around 50% with the persistence of the proatlantal artery. It usually corresponds to the V3 segment of the VA, originates from the internal or the external carotid artery and enters the skull through the foramen magnum [111][112][113]. Persistence of the hypoglossal artery is related to VA aplasia, with a prevalence of 0.02-0.1%. A persistent hypoglossal artery originates from the internal carotid artery and enters the skull via the hypoglossal canal [113,114]. The entrance in the skull via different foramina also represents a useful criterion in distinguishing the two arteries [115].

Clinical Relevance
Patients with variation in VA origin are exposed to an increased risk in AA or esophageal surgery, if the case is not previously documented. "Vertebral arteria lusoria" represents a VA originating from the AA, distal to the origin of the left SA, commonly with a retroesophageal course. This anatomical variant has been seldom reported in the literature [83,84,116,117]. Verin et al. reported the case of a lusoria VA entering the fourth cervical vertebra [118], and Meila et al. [52] reported the TF entrance of such a vessel at CThis condition was associated with life-threatening risk during esophageal surgery, as accidental injury to this vessel may lead to hemomediastinum or even neurological impairment [119].
Variation in the entrance into the TF should also be presurgical detailed, as it may expose the patient to risks during anterior neck surgery [51] or dissection during neck rotation [120].
Ignoring the VA anatomical variability could lead to its injury and life-threatening hemorrhage during anterior neck surgery, for instance, thyroidectomy or excision of a pharyngeal diverticulum, as the VA could be damaged while attempting ligation of the inferior thyroid artery [121].
Gluncic et al. [87] stated that atherosclerosis frequently affects the extracranial part of the VA. The origin of the VA is therefore predisposed to subsequent stenosis. Bernard and Dettori [122] presumed that an abnormal VA origin, distribution, or caliber favors cerebral disorders by altering cerebral hemodynamics. The study was conducted on two cases, so their hypothesis may not be generally applicable. Contrarily, Einstein et al. [33] stated that an aberrant LVA may have no vascular consequences, unless being compromised during AA surgery.
Tortuosity and hypoplasia of the VA have a considerable impact on posterior circulation infarctions and atherosclerosis [123,124]. The cause leading to tortuous vessels was speculated by Morris et al. [125] as being a thinner, fragile wall that displays as a meandering vessel. Another etiology proposed by Choi et al. [126] is that the VA adapts as the cervical spine develops a tortuous route due to osteoarthritis. Congenital anomalies of the craniovertebral junction [127] and trauma [128] were also suggested to determine tortuosity. Tortuosity is also linked with vascular vertigo [129] but only if one segment of the VA is affected; if no vascular risk factors exist, symptoms may not appear [130]. A twisted vessel is at higher risk of iatrogenic injuries, as it does not pursue a common course [131]. Thus, a pre-surgical imagistic recognition of the anatomical peculiarities of the VA is fundamental.

Study Limitations
Although, to our knowledge, this could be regarded as the most extensive metaanalysis study on the VA topic that comprehensively includes numerous anatomical aspects, we should admit that it has several limitations. First, as with any other meta-analysis, the present paper includes a great variety of studies, regardless of the patient's pathologies, geographic area, number of patients, study types, etc. To minimize this major heterogeneity, we added exclusion criteria, as the removal of papers with patients suffering from vascular disease or studies using less than 10 subjects. Second, considering the Origin and TF Entrance sections, the analyses included all the available studies on the specific anatomic variant. As the total number of evaluated arteries varied, the results are not comparable to each other. Considering the definition of a systematic review or a meta-analysis, we processed the available data in the literature; thus, the assessment of both crude and standardized prevalence was not attainable for each variant analysis. Our results are, thus, highly influenced by the reported results. To diminish any risk of affecting the outcomes of the present study due to biased results published by other authors, we filtered each article rigorously, applying the third exclusion criteria.

Small-Study Effect
The term "small-study effect" was proposed by Sterne et al. to express the observation that smaller studies typically produce greater effect sizes [132]. According to Yurasakpong et al., the small-study effect, which denotes publication bias, is a frequent issue, particularly in anatomical systematic reviews and meta-analyses [133]. Analyses with high standard errors (SE) and low sample size are more likely to report higher prevalence. Funnel plots of effect size against sample size were used to detect publication bias. The asymmetrical plots may suggest the potential impact of the small-study effect and publication bias on the results of the interested determinations.

Conclusions
The anatomy of the VA is extensively variable. The SA origin of the VA has a prevalence of 94.1%. The entrance of the VA at the sixth transverse foramen prevails in 92.0% of cases. Tortuosity (26.6%) and hypoplasia (7.94%) were also documented. Tortuosity in the transverse plane was the most frequent (30.1%), related to sagittal (22.8%) and coronal (16.6%) tortuosity. A straight V1 segment of the VA was prevalent in 45.2% of cases. Considering the VA dominance, LVA was dominant in 36.1% of cases, compared to RVA dominance, found in 25.3% of cases.

Conflicts of Interest:
The authors declare no conflict of interest.