Types of studies

Prospective or retrospective cross‐sectional studies that report the diagnostic accuracy of 18F‐FDG‐PET/CT in diagnosing lymph node involvement or bone metastases or lung metastases or a combination of these metastases in patients with confirmed RMS are eligible for inclusion. They must compare the results of the 18F‐FDG‐PET/CT imaging with the tests described as reference standards (as described below). Figure 1 shows the general criteria for considering studies for this review. Studies should report sufficient data to construct (part of) a 2×2 table, so the absolute number of true positives, false positives, true negatives, false negatives, or a combination of these have to be available from the data in the primary studies or to be obtainable from authors to re‐assess sensitivity and specificity. We will exclude \review articles, editorials or letters and case reports.

Figure 1

---

---

Criteria used to define eligible studies for this review. 18F‐FDG‐PET/CT: fluorine‐18‐fluorodeoxyglucose ‐ positron emission tomography/computed tomography; CT: computed tomography; MRI: magnetic resonance imaging; RMS: rhabdomyosarcoma.

Participants

Patients with histology confirmed RMS of any stage at first diagnosis. We will include studies with patients who are not eligible for this review (such as people with recurrence of RMS or other sarcoma types) if data for only the eligible participants are available.

Index tests

18F‐FDG‐PET/CT scans.

Target conditions

Newly diagnosed RMS with:

• bone metastases;

• lung metastases;

• nodal involvement;

• any combination of the above.

Reference standards

The most optimal reference standard for suspected distant metastases and lymph node involvement in RMS patients would be confirmation by histopathology obtained by biopsy. For both ethical and practical reasons, this cannot be done for every suspected lesion.

When biopsy results are not available, the results of the 18F‐FDG‐PET/CT should be compared with the judgement from a multidisciplinary tumour board, where experts have the knowledge of a patient's clinical findings, results from conventional imaging and histological data. Clinical follow‐up and imaging follow‐up could also be used to support the final diagnosis of nodal involvement, and bone and lung metastases (see Figure 1). In general, after nine weeks of chemotherapy, tumour response is evaluated with imaging including an X‐ray of the thorax.

• Bone and bone marrow involvement

A whole body 99m‐Tc skeleton scintigraphy and bilateral bone marrow aspirates and trephine biopsies are performed to identify bone metastases and bone marrow involvement. When possible, in case of doubt a biopsy is performed.

• Lung metastases

Lung metastases are detected by chest CT scan. In most patients, an X‐ray of the thorax will also be performed.

Pulmonary metastatic disease is defined as one or more pulmonary nodules of 10 mm or more of diameter or two or more well‐defined nodules of 5 mm to 10 mm diameter, in the absence of another medical explanation. In case of doubt or several small (less than 5 mm)
nodules, a multidisciplinary tumour board decides whether a biopsy is indicated to confirm the diagnosis.

• Nodal involvement

The presence of locoregional nodal involvement is evaluated using MRI and ultrasound. In case of doubt, a biopsy is performed. In addition to such conventional imaging modalities, for upper and lower limb tumours, it is highly recommended to have surgical evaluation of axillary (for upper limb tumours) or inguinal (for lower limb tumours) nodes, even if nodes are clinically or radiological normal.

Electronic searches

We will search two electronic databases: MEDLINE in PubMed (from 1966 to present) and EMBASE in Ovid (from 1980 to present).

Appendix 1 and Appendix 2 show the search strategies for the different electronic databases (using a combination of controlled vocabulary and text words).

Searching other resources

The review authors will locate information about studies not indexed in MEDLINE and EMBASE, either published or unpublished, by handsearching the reference lists of relevant articles and review articles. The review authors will also contact the authors of the included studies and other experts in the field of RMS for information about any ongoing or unpublished studies. The review authors will also scan conference proceedings electronically if available and otherwise by handsearching; we will search the International Society for Paediatric Oncology (SIOP), the American Society of Pediatric Hematology/Oncology (ASPHO), the Connective Tissue Oncology Society (CTOS), the American Society of Clinical oncology (ASCO) and the European Musculo‐Skeletal Oncology Society (EMSOS) (all from the five years up to the search date). In EMBASE, we will use the search fields conference publication (cg) and conference information (cf) in combination with Emtree terms and text words as mentioned in Appendix 2.

Selection of studies

After employing the search strategy described previously, two review authors will independently identify studies meeting the inclusion criteria. We will obtain in full text any study that seems to meet the inclusion criteria on the grounds of title, abstract or both, for closer inspection. Two review authors will independently undertake full‐text article screening. Only full‐text studies that fulfil all predefined criteria for considering studies for this review will be eligible for inclusion. We will clearly state reasons for exclusion of any study considered for the review. We will resolve disagreements during both initial selection and definite selection by consensus. If this is impossible, we will achieve final resolution using a third‐party arbitrator. We will include a flow chart of the selection of studies in the review.

Data extraction and management

Two review authors will independently perform data extraction using a predefined data extraction form. We will extract data on the following items:

• article: author, year of publication (or presentation), journal (or conference);

• study population: age at diagnosis, sex, histology (ARMS/ERMS), fusion status (PAX3/7‐FOXO1), primary tumour site, IRS group (I, II, III), nodal status, metastasis status (bone, lung, other), number of participants (including number eligible for the study, number enrolled in the study, number receiving the index test and reference standard, number for whom results are reported in the 2×2 table, reasons for withdrawal);

• index test; 18F‐FDG‐PET/CT scan including the system and protocol used and the definition of an 18F‐FDG‐PET‐positive lesion. Interpretation blinded to reference standards;

• conventional imaging modalities used (MRI, CT, or both of the primary site, chest CT‐scan, chest X‐ray, radionucleotide bone scan, cranial spinal MRI, ultrasound abdomen, CT and or MRI abdomen);

• reference standard: description of the reference standard used. Verification of findings by: biopsy of suspected lesions or judgement from an interdisciplinary tumour board (based on combination of clinical findings, results from conventional imaging, additional biopsy and follow‐up;

• study design: basic design of the study (prospective cohort or historical cohort with data collection based on medical records or case‐control study), time span between index test and reference test, treatment between index test and reference test;

• data for the 2×2 table: true positive, false positive, true negative and false negative rates or, if not available, relevant parameters (sensitivity, specificity or predictive values) to reconstruct the 2×2 table.

We will pilot the data extraction form using two studies.

When data are missing in a published report, we will attempt to contact the authors for the missing information. In case of disagreement, we will re‐examine the abstracts and articles and undertake discussion until we achieve consensus. If not possible, we will achieve final resolution using a third‐party arbitrator.

Assessment of methodological quality

Two review authors will independently assess each included study for methodological quality. For this, we will adapt a four‐domain tool from QUADAS‐2 (Whiting 2011). We will adapt this tool to our review; it comprises the following domains;

• participant selection;

• index test;

• reference standard;

• flow of participants through the study and timing of both the index text and reference standard (flow and timing).

For each domain, we will classify the risk of bias and concerns about the applicability of study findings as low, high or unclear. See Table 1.

For example, in domain 'Participant selection', we will evaluate whether a consecutive or random enrolment of participants has taken place. Some studies may have performed 18F‐FDG‐PET/CT solely in participants with unclear results obtained with standard tests, which could be a potential bias.

We will resolve discrepancies between review authors by consensus. If this is not possible, we will seek final resolution using a third‐party arbitrator.

We will summarize the methodological quality in the text, a graph and tables.

Statistical analysis and data synthesis

We will perform a participant‐based analysis of the data. We will analyze data of the three separate outcomes (nodal involvement, and lung and bone metastases) separately. We will use the data from the 2×2 tables (consisting of true positives, false positives, true negatives and false negatives) to calculate sensitivity and specificity for each study and each test. We will generate a paired forest plot showing estimates of sensitivity and specificity together with 95% confidence intervals. Such a forest plot provides a visual impression of the precision by which sensitivity and specificity have been measured in each study as well as an indication of the amount of variability in these parameters across studies. We will also plot pairs of sensitivity and specificity from each study in receiver operating characteristic (ROC) plots of sensitivity versus 1 minus specificity. We will provide likelihood ratios estimates. We will use a bivariate random‐effects approach to meta‐analyze the pairs of sensitivity and specificity from each study (Reitsma 2005). The reason for this is that we do not expect a cut‐off variability for the index test. We expect most studies will use a positivity cut‐off corresponding to maximum standardized uptake value (SUV_max) greater than 2.5 or a positivity criterion based on the surrounding background activity. The bivariate random‐effects approach is an appropriate framework to analyze such data as it correctly deals with: imprecision by which sensitivity and specificity are measured within each study; variation beyond chance in sensitivity and specificity between studies and any correlation that might exist between sensitivity and specificity.

We will perform analyses using STATA software.

Investigations of heterogeneity

When assessing study results, we will consider methodological and clinical sources of heterogeneity as well as variation in the criteria used to define a positive test result. Anticipated sources of heterogeneity include 18F‐FDG‐PET protocol (e.g. FDG dose), participant population (e.g. percentage of alveolar histology) and reference standard (biopsy confirmed or not). We will investigate heterogeneity using meta‐regression, that is, by incorporating covariates in the bivariate model. First, we will assess overall effects of covariates by likelihood ratio test comparing the bivariate model with and without the covariate, using a P value of less than 0.05 to denote significance. Then, with a significant likelihood test result, we will assess effects of covariates on sensitivity and specificity separately by testing the significance of the change in ‐2log‐likelihood of the model with or without corresponding terms.

We will assess the effect of such study characteristics by adding covariates in the bivariate models.

Sensitivity analyses

We will perform a sensitivity analysis of included studies that are at low risk of bias regarding participant selection, index test, reference standard, and flow and timing. We will omit studies that are at high or unclear risk of bias.

Assessment of reporting bias

We will undertake no formal assessment of reporting bias. However, we will highlight the possibility of reporting bias and interpret the results of any analysis cautiously.