Diagnostic accuracy of 1p/19q codeletion tests in oligodendroglioma: A comprehensive meta‐analysis based on a Cochrane systematic review

Abstract Codeletion of chromosomal arms 1p and 19q, in conjunction with a mutation in the isocitrate dehydrogenase 1 or 2 gene, is the molecular diagnostic criterion for oligodendroglioma, IDH mutant and 1p/19q codeleted. 1p/19q codeletion is a diagnostic marker and allows prognostication and prediction of the best drug response within IDH‐mutant tumours. We performed a Cochrane review and simple economic analysis to establish the most sensitive, specific and cost‐effective techniques for determining 1p/19q codeletion status. Fluorescent in situ hybridisation (FISH) and polymerase chain reaction (PCR)‐based loss of heterozygosity (LOH) test methods were considered as reference standard. Most techniques (FISH, chromogenic in situ hybridisation [CISH], PCR, real‐time PCR, multiplex ligation‐dependent probe amplification [MLPA], single nucleotide polymorphism [SNP] array, comparative genomic hybridisation [CGH], array CGH, next‐generation sequencing [NGS], mass spectrometry and NanoString) showed good sensitivity (few false negatives) for detection of 1p/19q codeletions in glioma, irrespective of whether FISH or PCR‐based LOH was used as the reference standard. Both NGS and SNP array had a high specificity (fewer false positives) for 1p/19q codeletion when considered against FISH as the reference standard. Our findings suggest that G banding is not a suitable test for 1p/19q analysis. Within these limits, considering cost per diagnosis and using FISH as a reference, MLPA was marginally more cost‐effective than other tests, although these economic analyses were limited by the range of available parameters, time horizon and data from multiple healthcare organisations.


INTRODUCTION
Complete deletion of both the short arm of chromosome 1 (1p) and the long arm of chromosome 19 (19q) (1p/19q codeletion) is a chromosomal alteration that occurs in oligodendrogliomas, but to date, the best method to detect such deletions is unclear. The codeletion is thought to be an early event in oligodendroglioma tumourigenesis [1] and is thought to be a result of an unbalanced whole-arm translocation between chromosomes 1 and 19 with the loss of the resulting hybrid chromosome [2,3] (Figure 1). The combined presence of an IDH1 or IDH2 mutation and a 1p/19q codeletion is a diagnostic criterion for oligodendroglioma, IDH mutant and 1p/19q codeleted [8].
The diagnostic test algorithm of IDH-mutant gliomas has been streamlined in a recent consensus publication cIMPACt-NOW update 5 [9], recommending that 1p/19q testing is not required in IDHmutant astrocytic tumours with loss of nuclear ATRX expression.
Although this recommendation reduces the number of 1p/19q tests in IDH-mutant gliomas, the diagnosis of oligodendroglioma, IDH mutant and 1p/19q codeleted, central nervous system (CNS) World Health Organization (WHO) Grade 2 or 3 still requires the detection of an IDH mutation and a 1p/19q codeletion. The European guidelines recommend that 1p/19q status is evaluated to support a diagnosis of oligodendroglioma, IDH mutant and 1p/19q codeleted, and for prognosis, and that treatment decisions are based on the 1p/19q status [10][11][12]. Current guidance from the National Institute for Health and Care Excellence (NICE) (United Kingdom) recommends and the 2021 CNS WHO classification [13] mandates testing 1p/19q codeletion to identify oligodendrogliomas, and the adjuvant chemotherapeutic recommended after surgery for people with CNS WHO Grade 3 glioma varies according to 1p/19q status (NICE 2018) [14].
1p/19q status can be determined by several different methods, and there is no consensus regarding the optimal method. The two most common methods for routine diagnostic use are FISH-and polymerase chain reaction (PCR)-based loss of heterozygosity (LOH) assays [15]. In the 2017 UK Cytogenomic External Quality Assess- favouring 1p/19q codeletion after adjusting for age, extent of resection, IDH mutation and type of therapy [16]. Another systematic review and meta-analysis found that 1p/19q codeletion was associated with increased overall survival (HR 0.43; 95% CI 0. 35-0.53; 14 studies) [17], both in WHO low-grade (HR 0.45; 95% CI 0.30-0.68; 5 studies) and high-grade oligodendrogliomas (HR 0.41; 95% CI 0.31-0.53; 6 studies), and for astrocytic tumours (HR 0.52; 95% CI 0.36-0.75; 3 studies) and oligodendroglial tumours (HR 0.41; 95% CI 0.30-0.56; 9 studies) [17]. This review also observed no evidence of difference in the HR for overall survival between studies using two different techniques (PCR-based LOH and FISH) to assess the status of chromosomal arms 1p and 19q [17]. It is important to note that these studies were carried out before the current definition of oligodendroglioma, which now mandates the presence of an IDH mutation and a 1p/19q codeletion.
1p/19q codeletion can be absolute, that is, loss in the presence of the normal number of other chromosomes, or relative if it occurs in the presence of polysomy (when cells contain at least one more copy of a chromosome than normal) or polyploidy (when cells contain more than two sets of chromosomes) ( Figure 1B-D). Several studies have suggested that people with relative 1p/19q codeletions (deletions in the presence of polysomy or polyploidy) have a worse prognosis (progression free survival or overall survival) than people with absolute 1p/19q codeletions, with some studies suggesting that prognosis in patients with relative codeletions may be similar to that of people with no codeletion at all [4][5][6]18]. In all these studies, classification of polysomy occurred when more than 30% of nuclei had more than two 1q and 19p signals, as assessed by FISH ( Figure 1E). Although there are limitations to these studies, for example, non-standardised treatment, these findings suggest that diagnosing absolute deletions is more important. The Cochrane review focuses primarily on detection of absolute deletions and in diagnosing situations where one copy of 1p/19q has been lost and the other copy duplicated (also termed copy-neutral LOH). Combinations of chromosomal deletions in oligodendrogliomas and the corresponding signals in FISH are presented in a schematic representation in Figure 1.
In addition to the significant clinical implications associated with the diagnostic accuracy of techniques to diagnose 1p/19q codeletion

Key Points
• In a Cochrane review, we established the most sensitive, specific and cost-effective techniques for determining 1p/19q codeletion status.
• Fluorescent in situ hybridisation (FISH) and polymerase chain reaction (PCR)-based loss of heterozygosity (LOH) test methods were considered as reference standard.
• Next-generation sequencing and single nucleotide polymorphism arrays have high specificity.
• No difference in the hazard ratio for overall survival was found between studies using two different techniques, PCR-based LOH and FISH. status in oligodendroglioma patients, there are also significant potential resource implications regarding the accuracy of the test. The estimated costs associated with clinical care for a patient with glioma ranged between US$ 4755 and US$ 42,907, with reported costs converted into 2013 US $ using an exchange rate based on purchasing power parities [19]. It was also estimated that 55% of these costs were attributable to chemotherapy drugs, chiefly temozolomide. If these therapies can be targeted at those patients who will obtain the greatest benefit, this will make better use of limited healthcare resources.
This review will assess the sensitivity and specificity of any DNAbased techniques that can be used on tumour tissue to evaluate 1p/19q codeletion status directly involved when performing the different test methods. In addition, a cost-effectiveness model was developed to equate costs against the diagnostic performance of each of the diagnostic test methods.

METHODS
The protocol for the review was published in the Cochrane Database of Systematic Reviews [20], and the review was undertaken and reported following Cochrane's guidance (which is consistent with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses [PRISMA]) [21]. A more detailed account of the methods and results can be found in the full Cochrane publication [20].

Study eligibility
We included cross-sectional studies that use two or more tests to assess 1p/19q status in tumour tissue from the same set of people.
Studies needed to present either raw data or classified results for patients for at least two tests. Studies that reported only on concordance of test results were not included. Studies with data for just one person were excluded. For the integrated review of economic evidence, we sought cost and full economic evaluations (costeffectiveness analyses, cost-utility analyses and cost-benefit analyses) that had been conducted alongside any study designs or as part of a modelling exercise. Participants were adults (≥18 years old) with glioma. Studies in which participants were recruited on the basis of their 1p/19q codeletion status were excluded.

Study selection, data extraction and quality assessment
We used EPPI-Reviewer 4 (https://eppi.ioe.ac.uk) for processes of screening and selection of studies and for part of the data extraction [22]. Data were extracted and further analysed in Microsoft Excel. Two reviewers assessed risk of bias and applicability of the DTA studies using the QUADAS-2 tool [23] tailored to our review.
F I G U R E 1 Graphical representation of absolute and relative 1p/19q codeletions. In all parts of the figure, chromosomes 1 and 19 are presented in separate frames to visualise the combination of FISH signals. The 1p and the 19q probes are red, and the reference probes (1q and 19p) are green. The approximate labelling sites are indicated in the chromosomal schematics. An unrelated chromosome (2) is also shown, and appearances as FISH images on the bottom of each frame. (A) Cell with diploid set of chromosomes, with two red signals each, for chromosomal arms 1p and 19q, as well as two green signals each for chromosomal arms 1q and 19p. (B) Absolute 1p/19q codeletion in a diploid set of chromosomes. Loss of one red signal in chromosome 1p and in 19q and two green signals for each 1q and 19p. (C) Relative codeletion with example of polysomy of chromosome 19 and chromosome 2, which has been suggested to indicate a worse prognosis [4][5][6][7]. (D) 1p/19q codeletion in tetraploid cells, resulting in two red and four green signals for both 1p and 19q tests. (E) Complex deletion patterns as found in a small proportion of oligodendrogliomas, often associated with anaplastic histological types. In this example, there are diploid cells (left, 30%) triploid cells (centre, 30%) and tetraploid cells (right, 40%) Disagreements were resolved by consensus, with discussion with a third review author if necessary.

Index tests and target conditions
Studies using any DNA-based technique to determine 1p/19q status in tumour tissue were included, whereas studies using immunohistochemically detection of 1p/19q status, or studies assessing 1p/19q status from blood samples of imaging, were excluded. The target condition was an absolute 1p/19q codeletion, that is, in the absence of polysomy. As described in Table 1 a Narrative for these fields: All hypothetical scenarios assume that 31 people out of 100 with glioma will have a FISH-detected 1p/19q codeletion. Taking the example of CISH: Of these, 31 people will be given the correct positive result and 0 people will be given a false negative result; of the 69 people without the codeletion, 68 people will be given a correct negative result and 1 people will be given a false positive result.
b Narrative for these fields: All hypothetical scenarios assume that 31 people out of 100 with glioma will have a PCR-based LOH-detected 1p/19q codeletion. Taking the example of NGS: Of these, 31 people will be given the correct positive result and 0 people will be given a false negative result. Of the 69 people without the codeletion, 68 people will be given a correct negative result and 1 people will be given a false positive result.
T A B L E 1 (Continued) a Narrative for these fields: All hypothetical scenarios assume that 31 people out of 100 with glioma will have a FISH-detected 1p/19q codeletion. Taking the example of CISH: Of these, 31 people will be given the correct positive result and 0 people will be given a false negative result; of the 69 people without the codeletion, 68 people will be given a correct negative result and 1 people will be given a false positive result.
b Narrative for these fields: All hypothetical scenarios assume that 31 people out of 100 with glioma will have a PCR-based LOH-detected 1p/19q codeletion. Taking the example of NGS: Of these, 31 people will be given the correct positive result and 0 people will be given a false negative result. Of the 69 people without the codeletion, 68 people will be given a correct negative result and 1 people will be given a false positive result.

Statistical analysis and data synthesis
For analysis with each of the respective reference standards (FISH or PCR-based LOH tests), we performed bivariate meta-analyses of the sensitivity and false positive rate (1Àspecificity) of each index test, assuming binomial likelihoods for the number of 'true positive' and 'true negative' test results (2 Â 2 table) [24,25]. This approach allows for heterogeneity in sensitivity and specificity across studies and for between-study correlation in these measures. In our main analyses, we assumed that this between-study correlation and the standard deviation (heterogeneity) parameters were shared (i.e., identical) across tests. For studies comparing more than one test with the reference standard, multiple 2 Â 2 tables were derived. into 2020 GBP using the EPPI-Centre Cost Converter [29].

Deviation from protocol
We had planned to perform a latent class analysis of all available data.
We did not do this due to the complex structure of the data (with

Search results and included studies
Using the search methodology (Section 2), 5427 records were identified, and after removing duplicates, 3010 records were screened at title and abstract; 237 records were selected for full-text review, and 53 studies (in 78 publications) met the inclusion criteria. Assessments of risk of bias were mixed, due largely to lack of information about procedures in the study reports. The main issue of applicability was that many studies included only patients with specific subtypes of glioma.

Presentation of study findings
A network plot illustrates comparisons of test methods that were made among the included studies ( Figure 3). A summary of the study findings and meta-analysis results is presented in

Comparison of studies with FISH as reference standard
From the included studies that performed FISH ( Figure 7B) and at least one other test that was not a FISH variant, we created 41 crossclassified 2 Â 2 tables, with FISH as the reference standard (Table 1).   (Figures 4-7). The main results from the bivariate meta-analysis model indicate that sensitivity and specificity were generally high, though with wide credible intervals for most tests, and some results are based on very small numbers of patients, such as the result for mass spectrometry, which is based on a single study of 10 people.
Our GRADE assessments for all tests were either 'low' or 'very low'.

Comparison of studies with PCR-based LOH as reference standard
From the included studies that performed PCR-based LOH ( Figure 7B) and at least one other test that was not a PCR-based LOH  [49]. The top of the figure indicates a graphical representation of chromosome 1 (adapted from the GRCh38/hg38 Assembly). For legend to symbols, see Figure 4 variant, we created 32 cross-classified 2 Â 2 tables, treating PCRbased LOH as a reference standard (  (Figures 4-7). Results from the main bivariate meta-analyses are again based on very low numbers of patients. A poor estimate of sensitivity for G banding/karyotyping is based on a single study in which none of 13 PCR-detected 1p/19q codeletions were identified.
Our GRADE assessments for all tests were either 'low' or 'very low'.

Results from economic model
The results for the base case of the economic model are summarised in Table S1 (FISH as reference standard) and  Tables S1 and S2 must be interpreted with caution.

Sensitivity analysis
The results of the sensitivity analysis are displayed in Tables S1 and S2. The cost-effectiveness was compared with a number of different thresholds of societal willingness to pay (WTP) for the three outcomes: (i) cost per true positive, (ii) cost per true negative or (iii) cost per case detected. These thresholds ranged from GBP (£) 0 (i.e., the decision is made on cost alone, and the test with the lowest cost would always be considered the most cost-effective) to £10,000 (i.e., the amount willing to be paid for an additional unit of effect such as an additional true positive detected  [75][76][77][78][79][80]. The illustration in Figure 1 shows the location of probes on the 1p chromosome. Our test accuracy results confirm previous studies [17], showing that there is no difference in the HR for overall survival between studies using two different techniques, PCR-based LOH and FISH.
A technique of increasing importance is methylation array profiling. It is primarily used to establish epigenetic profiles of brain tumours, but the array data also generate a copy number profile with the added benefit of visualising chromosomal aberrations including 1p/19q codeletion [75,76,[78][79][80]. This has been reported in two comparative studies [81,82]. Future research can focus on deriving more longitudinal data to inform future economic evaluation studies assessing the long-term health costs and consequences of such strategies.

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no new data were created or analysed in this study.