Abstract
BACKGROUND: Low back pain is common worldwide. MR imaging may identify extraspinal findings that are not related to the proposed clinical question. The prevalence of extraspinal incidental findings and their clinical significance has not been well-established.
PURPOSE: This review aimed to evaluate the prevalence of extraspinal findings on MR imaging of the lumbar spine in adults and the prevalence of clinically significant incidental findings.
DATA SOURCES: A systematic search of MEDLINE and EMBASE was performed, including studies published before June 14, 2023.
STUDY SELECTION: Studies presenting a prevalence of extraspinal findings in patients 16 years of age or older were included.
DATA ANALYSIS: A random effects meta-analysis was used to generate composite prevalence measures of extraspinal findings, patients with extraspinal findings, and clinically significant findings.
DATA SYNTHESIS: Sixteen studies were included in this meta-analysis, with a total of 19,593 patients and 6,006 extraspinal incidental findings. The overall prevalence of extraspinal findings was 19.9% (95% CI, 11.1%–30.7%). Overall, 26.7% of patients had an extraspinal finding identified (95% CI, 14.8%–40.6%). The most common subgroup of extraspinal findings was genitourinary findings in males (27.1%; 95% CI, 25.6%–28.8%). Data from 8 studies demonstrated the prevalence of clinically significant findings at 5.4% (95% CI, 3.2%–8.1%).
LIMITATIONS: Retrospective populations with small numbers of participants in clinically relevant subgroups may result in heterogeneity and imprecision within composite outcome measures.
CONCLUSIONS: Extraspinal findings are common, detected in more than one-quarter of patients. Five percent of findings are clinically significant and may require further action.
ABBREVIATIONS:
- C-RADS
- CT Colonography Reporting and Data System
- ESIF
- extraspinal incidental finding
- LBP
- low back pain
Low back pain (LBP) is a commonly reported symptom globally and, owing to an ever-growing and aging population, prevalence is rapidly increasing. The lifetime prevalence of LBP internationally is approximately 80%–85% and accounts for 7.4% of global years lived with disability, the highest of any disease.1,2 MR imaging of the lumbosacral spine is the recommended technique for investigation of LBP.
Although most causes of LBP will represent neuromusculoskeletal pathologies, other findings may be identified on imaging. Extraspinal incidental findings (ESIFs) on MR imaging of the lumbar spine are those detected on a study which does not relate to the clinical question of LBP. ESIFs detected on lumbar spine imaging will commonly be found within the abdomen or pelvis, relating to the gastrointestinal, genitourinary, and gynecologic systems. The detection of these findings will vary depending on the imaging protocols, namely, FOV sampled, the use of saturation banding, and imaging sequences acquired. With the development of modern imaging software and improvement of imaging techniques and quality, diagnostic opportunity has increased.3 Subsequently, it follows that with more MR imaging examinations being performed, more ESIFs will be identified. ESIFs pose a challenge for clinicians. Some of these findings may allow earlier diagnosis and treatment of a significant pathology. However, some findings may not be of clinical significance or remain indeterminate. Insignificant incidental findings may lead to unnecessary follow-up diagnostic tests. Patients may be exposed to unnecessary procedures that may have inherent risks, in addition to the psychological impact of diagnostic uncertainty in the interim period. Insignificant incidental findings may culminate in an unnecessarily increased burden on global health systems.
This systematic review aimed to determine the prevalence of ESIFs in adults who underwent MR imaging of the lumbar spine. The prespecified secondary outcomes of this review were to determine the prevalence of clinically significant ESIFs and the prevalence of ESIFs stratified by age, sex, and body region.
MATERIALS AND METHODS
The protocol for this systematic review and meta-analysis was published on the PROSPERO database (CRD42023430740). The review has been reported in accordance with published Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.4
Search Strategy
Published articles and abstracts were identified following comprehensive searches of the MEDLINE and EMBASE databases using the Ovid platform. The search strategy (Online Supplemental Data) used the following search terms (and related synonyms): “spine,” “MR imaging,” “incidental,” and “extraspinal.” No date restrictions were applied. Search results were updated to include studies published before June 14, 2023.
Studies were considered eligible for inclusion in the systematic review if they satisfied the following criteria: 1) patients 16 years of age or older, 2) patients referred for lumbar spine MR imaging with LBP, and 3) studies reporting a prevalence of ESIFs. Studies were considered ineligible if prevalence data could not be extracted, the patients were derived from disease cohorts, or the article was published in a language other than English. Two reviewers (P.J.B. and J.E.C.) independently screened abstracts and identified full-text articles for review. All reference and citation lists of included full-text articles were interrogated using Google Scholar to identify further potential records. In the event of any discrepancies, these were resolved through consensus agreement between reviewers.
Data Extraction
Three reviewers (P.J.B., E.G., and A.E.K.) independently extracted data from each included study. Study characteristics extracted included the following: 1) study design, 2) study population, 3) setting, 4) geographic location, 5) recruitment period, 6) sample size, 7) imaging protocol, 8) exclusion criteria, 9) interpretation definition, 10) extraspinal finding definition, 11) clinically significant finding definition, 12) number of patients with extraspinal findings, 13) number of extraspinal findings, 14) number of clinically significant findings, and 15) number of patients who underwent imaging follow-up. The number of extraspinal findings was stratified by body systems, when outlined by the original study. Participant characteristics extracted included sex, age, and ethnicity.
Quality Assessment
Two reviewers (P.J.B. and J.E.C.) independently performed a risk of bias assessment of all included studies using a modified risk of bias assessment tool designed for prevalence studies (Online Supplemental Data).5 Studies were scored across 9 domains with a total score of 9. Studies with a low risk of bias had a score of 0–3; moderate risk, 4–6; and high risk, 7–9. Discrepancies between reviewers were resolved through consensus discussion.
Statistical Analysis
The primary outcome was a composite of the prevalence of extraspinal findings identified on MR imaging of the lumbar spine. Secondary outcomes included the percentage of patients with extraspinal findings, extraspinal findings within each body system, and the number of clinically significant findings identified. Statistical analyses were performed using STATA, Version 17.0 (StataCorp). The STATA program metaprop was used to perform a random effects meta-analysis of proportions extracted from each study.6 The random effects model was used on the basis of the assumption of heterogeneity among studies. The Freeman-Tukey double arcsine transformation was used to stabilize the variances of the reported proportions, with confidence intervals calculated using the Wald method. Univariate and multivariable (sex- or age-adjusted) meta-regression was performed to explore the effects of participant characteristic on the prevalence of extraspinal findings. Heterogeneity across studies was estimated using the I2 statistic. I2 values of >75% suggested substantial heterogeneity. Publication bias and small study effects were evaluated through visual inspection of the Doi plot and Egger test.7
RESULTS
Following the database search, 782 articles were identified and screened for eligibility. Of these, 26 full-text articles were assessed for inclusion, and 6 studies were selected.8⇓⇓⇓⇓-13 Twenty studies were excluded, on the basis of data presented being insufficient to meet the aims of the review (n = 16), patients included derived from a known disease cohort (n = 2), and the study population not including imaging of the lumbar spine (n = 1). One study was excluded because it was not published in English. Following a search of reference and citation lists, 16 full-text articles were screened for eligibility and 10 of these were selected for inclusion.14⇓⇓⇓⇓⇓⇓⇓⇓-23 In total, 16 studies were included in this review (Online Supplemental Data).8⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓-23
Study Characteristics
The characteristics of the 16 studies are summarized in the Online Supplemental Data. Eight studies were from Europe,10,13,15⇓⇓-18,20,22 5 were from Asia,9,11,14,19,21 2 were from the United States,12,23 and 1 was from Africa.7 All were retrospective cross-sectional studies. All 16 studies reported the number of ESIFs.8⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓-23 Ten studies reported the number of patients with ESIFs.8⇓⇓⇓⇓-13,15,19,22,23 Eight studies reported the number of clinically significant ESIFs.10⇓⇓-13,15,18,22,23 Studies that provided data for body system subgroup analysis can be found in the Online Supplemental Data. Four studies used a minimum imaging protocol consisting of axial T2WI, sagittal T1WI, and T2WI sequences.10,13,14,21 Ten studies included additional sequences as part of their imaging protocol.8,9,11,12,15,17,19,20,22,23 Two studies did not report the imaging protocol used.16,18 Six studies operationalized the nomenclature for extracolonic findings from the modified CT Colonography Reporting and Data System (C-RADS) classification to report clinically significant findings.10,12,13,15,18,22 The C-RADS reporting method numbers both colonic (C) and extra-colonic (E) findings. Findings labeled E3 and E4 are defined as “likely unimportant finding, incompletely characterized– subject to local practice and patient preference, work-up may be indicated” and “potentially important finding,” respectively.24 One study defined clinically significant findings as “findings with MR appearance characteristics of significant diseases, indeterminate lesions requiring further assessment by clinical lab/correlation, or findings requiring further imaging studies or histopathological or surgical confirmation after a discussion with the relevant physician.”11 One study defined clinically significant findings as the detection of an abdominal aortic aneurysm of ≥3 cm.23
Participant Characteristics
A summary of the participant characteristics is presented in the Online Supplemental Data. The age of participants was between 16 and 95 years. Eleven studies reported the distribution of sex in their total cohorts, with 44% of participants being males. Participant ethnicity was not recorded in any of the studies. Common reasons for participant exclusion in studies included a history of recent trauma, known malignancy, previous spinal surgery, or nondiagnostic images acquired.
Prevalence of All Extraspinal Findings
The total prevalence of all ESIFs identified from MR imaging of the lumbar spine, comprised 19,593 participants from 16 studies, was 19.9% (95% CI, 11.1%–30.7%) (Table 1). The prevalence of ESIFs in females from 3 studies was 55.1% (95% CI, 12.7%–93.2%). The highest proportion of ESIFs was identified in the genitourinary system in males (Table 2). Incidental genitourinary findings were more common in males compared with females 27.1% (95% CI, 25.6%–28.8%) versus 17.5% (95% CI, 16.3%–18.7%). The prevalence of gastrointestinal and urinary tract findings was 2.0% (95% CI, 0.3%–5.2%) and 8.7% (95% CI, 4.7%–13.8%), respectively. Studies that included additional sequences as part of their imaging protocol did not demonstrate a higher reported prevalence of findings (22.0%; 95% CI, 13.2%–32.4%) compared with studies that performed the minimal imaging sequences (24.5%; 95% CI, 2.9%–57.9%).
Estimated prevalence for extraspinal findings, patients, and clinically significant findings
Estimated prevalence of extra-spinal findings by body region
Prevalence of All Patients with Extraspinal Findings
The total prevalence of all patients with ESIFs identified on MR imaging of the lumbar spine, comprised of 14,992 participants across 10 studies, was 26.7% (95% CI, 14.8%–40.6%) (Table 1). The prevalence in females from 3 studies was 51.0% (95% CI, 2.9%–97.8%). The prevalence in patients older than 40 years of age from 4 studies was 16.2% (95% CI, 4.8%–32.7%). There was no difference in the prevalence of patients with ESIFs when stratified by sex or age.
Prevalence of Clinically Significant Findings and Imaging Follow-Up
The total prevalence of clinically significant ESIFs, comprised of 12,778 participants from 8 studies, was 5.4% (95% CI, 3.2%–8.1%) (Table 1). There was no difference in the prevalence of clinically significant findings in patients stratified by sex. No studies reported the prevalence of clinically significant findings in patients younger than 40 years of age. Two studies reported the prevalence of clinically significant findings in patients 40 years of age or older (10.0%; 95% CI, 7.8%–12.4%). Only 1 study reported the number of patients who underwent imaging follow-up; 32.6% of patients with extraspinal findings required further imaging.8
Meta-Regression Analysis
Following univariate meta-regression, no predefined characteristics (sex, age, or continent) demonstrated a significantly higher proportion of extraspinal findings (Online Supplemental Data). In univariate analysis, the risk difference for males was −18.4% (95% CI, −61.8%–25.0%). The risk difference for patients 40 years of age and older was 15.5% (95% CI, −32.6%–63.4%). These findings persisted in the multivariable analysis performed, with no predefined characteristic demonstrating a statistically significant difference in the percentage risk difference.
Study Quality and Small Study Effects
Following an assessment of risk bias using an adapted tool for prevalence studies, total scores ranged from 1 to 7 (Online Supplemental Data). All studies scored in domain 2 due to single-center study design. All studies scored zero in domain 9, given that sufficient data to extract prevalence proportions formed part of the inclusion criteria for this systematic review. Eight studies were assessed as a low risk of bias.9⇓-11,13,15,17,18,20 Seven studies were evaluated as having a moderate risk of bias,8,12,14,16,19,21,23 with only 1 study evaluated as having a high risk of bias.22 Post hoc analysis showed that the prevalence of ESIFs in the study evaluated as having a high risk of bias was 36.3% (95% CI, 30.8%–42.1%) versus 27.6% (95% CI, 12.9%–45.2%) in studies evaluated as having a low risk of bias. Therefore, no studies were excluded from analysis on the basis of risk of bias assessment. Visual inspection of the Doi plot demonstrated lateralizing asymmetry with a LFK index value of −1.71, suggesting small study effects (Egger test, P = .059) (Online Supplemental Data).
DISCUSSION
In summary, this systematic review consisted of 16 studies including 19,593 participants with 6,006 extraspinal findings identified on MR imaging of the lumbar spine. In 10 studies, including 14,992 patients, 5,215 patients had extraspinal findings. The overall prevalence of an extraspinal finding was 19.9%, and the prevalence of a patient having an extraspinal finding was 26.7%. Clinically significant findings were identified in 5.4% of participants. The prevalence of ESIFs was most pronounced in patients 40 years of age or older.
To the best of our knowledge, this is the first systematic review and meta-analysis to assess the prevalence of ESIFs on MR imaging of the lumbar spine in adults. Incidental findings on MR imaging of the lumbar spine in asymptomatic pediatric patients have previously been reported and subsequently included in an umbrella review of incidental findings across multiple imaging modalities.25,26 Our reported composite prevalence is similar to the prevalence reported in the pediatric population and the prevalence across other imaging in the umbrella review. Studies that were excluded from our review because they included some patients younger than 16 years of age also reported similar prevalence figures.27,28 When one considers multitechnique evidence, the prevalence of extraspinal findings on outpatient lumbar spine CT in a cohort of 400 adults was 40.5%, with 14.8% of patients having indeterminate/clinically significant findings, which required further clinical assessment or follow-up imaging.29 The higher percentage of ESIFs on CT may be explained by technical factors relating to image acquisition (for example, contiguous section interval and section thickness) and the differential sensitivity of detecting findings with fat or air densities through this technique.
The findings of this review have both clinical and research implications. Our results demonstrate that extraspinal findings are common, with more than one-quarter of examinations having identified ESIFs. Our analysis showed that there was a pronounced difference in the percentage of genitourinary findings in males compared with females. The proportion of genitourinary incidental findings in males was calculated from 2 studies.10,15 These studies included renal cysts, some of which were classified as clinically important. Our analysis is similar to previously reported studies evaluating the prevalence of renal cysts, which demonstrated more prevalent cystic disease in males.30,31 The clinical implications pertain to the responsibility of the radiologists to identify these ESIFs and, if appropriate, notify referring clinicians of clinically significant findings. Radiologists should consider revising routine search patterns to place a greater emphasis on the genitourinary tract in males. This change could be made by including such review areas in standard reporting templates. Five percent of findings were deemed clinically significant. This translates to a sizeable proportion of examinations that may include actionable findings that require further work-up. Both radiologists and referring clinicians should be equipped to appropriately manage detected extraspinal findings and manage the uncertainty of suspected significant findings. This recommendation is important to minimize overlooking significant diagnoses but also to avoid unnecessary further imaging or procedures, which may bring their own physical risk and elevated patient anxieties.
This review has also highlighted the relative paucity of evidence when stratifying for participant characteristics. Only 1 study included in this review reported the number of patients who underwent follow-up on the basis of the extraspinal findings detected. There is a need for further observational studies designed to capture the prevalence and impact of extraspinal findings in clinically relevant subgroups. Another future research question is the potential impact of technical imaging factors on the prevalence of ESIFs. Technical differences, namely, FOV, section thickness, and the use of saturation bands are likely to influence the detection of ESIFs. Inclusion of saturation banding will limit assessment of the prevertebral structures. This phenomenon is explained by the reduced detection of lymphovascular disease on MR imaging of the lumbar spine compared with CT.29,32 Anecdotally, localizer images could identify ESIFs, and their utility could be quantified. The included studies did have variance in their predetermined imaging protocols. Studies that included additional imaging sequences did not report a higher proportion of ESIFs; however, this observation seems counterintuitive.
Strengths and Limitations
This review has multiple strengths. First, the review protocol was preregistered and the review used a robust and comprehensive search strategy to exhaustively ensure that all relevant studies from the published literature were included. Second, we intentionally included studies without a prespecified definition of extraspinal findings. We thought that during data extraction, extraspinal findings could be pragmatically identified among all documented findings. This idea has allowed us to generate stratified analyses for clinically relevant patient characteristics, body regions, and clinically significant findings. Third, we excluded studies reporting data from pre-existing disease cohorts because this would introduce bias.
One limitation of this systematic review is that only 2 databases were included. However, this has been mitigated by using a third database when reviewing the reference and citation searches. Other limitations of this review arise from the included studies. First, most studies are from single centers and demonstrate marked heterogeneity with small study effects, limiting the generalizability at a population level. However, there was global geographic coverage from the studies, and this will likely mitigate the former. Second, the estimated proportions, particularly in subgroup analysis and univariate analysis, are limited because few studies reported extraspinal findings stratified by clinically relevant subgroups. This may result in some imprecision in the composite prevalence calculations. Third, there was heterogeneity in the interpretation and definition of imaging findings across studies, with multiple subspecialty radiologists reporting studies. This will introduce subjective bias in the reporting rate across studies. Last, there were differences in the reporting of an extraspinal finding and a clinically significant finding, with some authors having lower or higher thresholds. For example, in 1 study, pelvic free fluid was documented as a positive finding; this influenced the prevalence of findings in females.10 Another study only reported the presence of an abdominal aortic aneurysm of ≥ 3 cm as clinically significant, which will introduce bias into the vascular subgroup.23 However, there was a relative consensus among the remaining studies on how to document a finding as clinically significant, improving the external validity, with most studies adopting the modified C-RADS classification.
CONCLUSIONS
ESIFs detected on MR imaging of the lumbar spine are common and were reported in approximately one-quarter of studies. Clinically significant findings were reported in 5%. Incidental genitourinary findings were more prevalent in males compared with females. Both referring clinicians and reporting radiologists should be aware of the implications of incidental findings and how to manage them in their routine clinical practice. There is a relative paucity of evidence in some patient subgroups, and future population-based studies should address this.
ACKNOWLEDGMENTS
We thank Dr Leo Alexandre, for his thorough teaching in systematic review methods, without whom this review would not be possible.
Footnotes
Disclosure forms provided by the authors are available with the full text and PDF of this article at www.ajnr.org.
References
- Received June 26, 2023.
- Accepted after revision October 18, 2023.
- © 2024 by American Journal of Neuroradiology
Jump to section
Related Articles
Cited By...
- No citing articles found.