Imaging modalities for characterising focal pancreatic lesions
Abstract
This is a protocol for a Cochrane Review (Diagnostic test accuracy). The objectives are as follows:
To determine and compare the diagnostic accuracy of various imaging modalities in detecting cancerous and precancerous lesions in patients with focal pancreatic lesions.
Background
A 'shadow’ identified in the pancreas on imaging may be variously described as a focal pancreatic lesion, a pancreatic mass, a pancreatic tumour, a pancreatic cyst, or a pancreatic nodule. This refers to focal lesions as opposed to diffuse changes of pancreas and includes solid and cystic lesions of the pancreas. In the western world, the prevalence of focal pancreatic lesions is approximately 1.2% (Spinelli 2004) and is increasing steadily (by approximately 8%) each year with smaller and asymptomatic lesions being identified more frequently (Gaujoux 2011). An incidental pancreatic lesion is one that is detected in the pancreas of a patient who undergoes radiological investigations for unrelated medical conditions (Sachs 2009). Such asymptomatic incidental lesions represent 55% to 60% of pancreatic tumours (Gaujoux 2011; Spinelli 2004). Some focal pancreatic lesions may be associated with symptoms depending upon their nature. The symptoms of pancreatic cancer, which generally refers to pancreatic adenocarcinoma, may include obstructive jaundice (yellowish discolouration of the skin and the whites of the eyes with dark urine and pale stool due to blockage of bile duct (National Cancer Institute 2011a), a tube which transports the bile from the liver), loss of appetite, and pain (Holly 2004). The symptoms of pancreatic neuroendocrine tumours, some of which may be malignant, are related to the excessive secretion of hormones (by the tumour) such as insulin, glucagon, gastrin, somatostatin, and vasoactive peptide resulting in hypoglycaemia (decreased blood sugar), hyperglycaemia (increased blood sugar, a rare cause of diabetes), and gastrointestinal disturbances such as peptic ulcer and diarrhoea (Batcher 2011). The symptoms of chronic pancreatitis (chronic inflammation of the pancreas that can result in alteration in the structure and function of the pancreas) are pain (Braganza 2011) and those related to pancreatic insufficiency, which include steatorrhoea, malabsorption, vitamin deficiency, diabetes, or weight loss (Nair 2007). About 40% of patients with focal pancreatic lesions have chronic pancreatitis (Spinelli 2004). In the remaining 60% of patients with focal pancreatic lesions, the remaining pancreas is normal.
Focal pancreatic lesions can be benign (pancreatic cystadenoma, papillary cysts, lymphoepithelial cysts, simple cysts), precancerous (intraductal papillary mucinous neoplasm (IPMN) without invasive cancer, mucinous cystic neoplasm (MCN), benign neuroendocrine tumours), or cancer (IPMN with invasive carcinoma, cystadenocarcinoma, malignant neuroendocrine tumours) (Sachs 2009; Spinelli 2004). About 80% of benign lesions, 50% of precancerous lesions, and 20% of malignant lesions are asymptomatic (Spinelli 2004).
Surgical resection is generally considered as the only treatment that can achieve cure of pancreatic cancer. Worldwide, only 15% to 20% of patients with pancreatic cancers undergo potentially curative resection (Conlon 1996; Engelken 2003; Katz 2009; Michelassi 1989; Shahrudin 1997; Smith 2008). In the remaining patients, the cancers are not resected because of infiltration of local structures or because of disseminated disease. Early diagnosis of pancreatic cancer might enable resection of the pancreatic cancer before it is too late to resect. Pancreatic resection is a major surgery with approximately 1% to 25% risk of perioperative death reported worldwide (Conlon 1996; Katz 2009; Michelassi 1989; Shahrudin 1997; van Oost 2006). High‐volume centres show a lower perioperative mortality of less than 5% as compared to low‐volume centres, which are associated with a perioperative mortality of around 25% (van Oost 2006). Pancreatic resection is also associated with about 40% morbidity rate (van der Gaag 2010). Only 5% to 25% of patients survive for five years (Conlon 1996; Katz 2009; Michelassi 1989; Shahrudin 1997). Surgery is generally offered if there are features suggestive of precancerous or cancerous lesions (Lee 2005), although some clinicians prefer sequential follow‐up (by imaging) of precancerous lesions to surgical resection (Irie 2004). Surgery is offered when there is an increase in the size of the lesion in sequential imaging (Gaujoux 2011). Surgery is also offered when there is considerable uncertainty as to the nature of the lesion. In some ways, surgery can be considered as a diagnostic test for characterisation of the lesion and as a treatment for patients with cancerous and precancerous lesions. Histological confirmation of the lesion by biopsy is not generally performed because of the fear of dissemination of cancer cells.
There is no standard algorithm in the diagnosis or management of focal pancreatic lesions. The algorithm may vary from one centre to another and even within the same centre (Goh 2006; Gaujoux 2011). One possible diagnostic clinical pathway is shown in Figure 1. It is important to distinguish whether the focal pancreatic lesion is benign with no malignant potential so that unnecessary surgery and anxiety can be avoided. It is important to know whether the lesion is precancerous or cancerous so that an informed decision about the surgery can be made after weighing the benefits and harms. It is also important to differentiate the different types of malignancy since different malignancies carry different prognosis (Klempnauer 1995). Some surgeons follow the single test strategy making decisions based on the features of the lesion in a single test while others follow repeated testing (using the same imaging modality as before or by a different imaging modality). The optimal interval between the tests in the repeated testing strategy is not known.
Clinical pathway.
Ca 19‐9: carbohydrate antigen 19‐9;
CT: computerised tomogram;
EUS: endoscopic ultrasound;
MRI: magnetic resonance imaging;
PET: positron emission tomogram.
Target condition being diagnosed
Benign versus precancerous and cancerous lesions (including the type of cancerous lesion).
Index test(s)
Computerised tomography (CT) scan: this involves a series of X‐rays taken from different angles, which are then reconstructed using a computer (National Cancer Institute 2011a). Morphological features of the lesion, such as density, regularity of margins, vascularity, and the diameter of the pancreatic duct, are taken into account to characterise the lesion.
Magnetic resonance imaging (MRI): this involves the use of a powerful magnet to produce images of different tissues of the body. This is also called nuclear magnetic resonance imaging (NMRI) (National Cancer Institute 2011b). Similar features as those in CT scan are used to characterise the lesion.
Positron emission tomography (PET): this involves the use of a small amount of radioactive glucose (sugar) to differentiate between different tissues. It utilises the property that cancer cells often use more glucose than normal cells. It is also called PET scan (National Cancer Institute 2011c). Cancerous lesions appear as areas of increased uptake.
Endoscopic ultrasound: this involves the use of an endoscope, a camera introduced into the body cavities to view the inside of the body. An ultrasound (high‐energy sound waves) probe at the end of the endoscope is used to differentiate different tissues. This is also called endosonography and EUS (National Cancer Institute 2011d). Features such as echogenecity and regularity of margins are taken into account and are used to characterise the lesion.
EUS elastography: this measures the stiffness of the lesion. The stiffness of the lesion may be used to identify whether the lesion is benign or malignant (Iglesias‐Garcia 2010).
EUS‐guided biopsy: this is the removal of cells or tissues for examination by a pathologist. The pathologist may study the tissue under a microscope or perform other tests on the cells or tissue. There are many different types of biopsy procedures. The most common types include: (1) incisional biopsy, in which only a sample of tissue is removed; (2) excisional biopsy, in which an entire lump or suspicious area is removed; and (3) needle biopsy, in which a sample of tissue or fluid is removed with a needle. When a wide needle is used the procedure is called a core biopsy. When a thin needle is used the procedure is called a fine‐needle aspiration biopsy (FNAB) (National Cancer Institute 2011e). Because of the risk of dissemination from cancer, EUS‐guided biopsy is preferable to percutaneous (image‐guided) biopsy (Micames 2003). The examinations under the microscope may include the routine Haemotoxylin and Eosin stain for core biopsy and special staining for FNAB (Mehta 2010). In addition, immunocytochemistry and proteomic profiling to identify the presence of biomarkers in the tissue may be used in the diagnosis (Cui 2009; Mehta 2010). A positive core biopsy can confirm malignancy but a negative core biopsy cannot rule out malignancy. Cytology results are not quite as reliable as core biopsy since false‐positive cytology has been reported (Hancke 1984). Currently, biopsy in any form is generally avoided because of the risk of cancer spread (cancer seeding).
Alternative test(s)
There is no standard algorithm for assessing focal pancreatic lesions. CT scan is the most likely investigation used (Gaujoux 2011). If the incidental lesion is detected on CT scan, then CT scan can be the only investigation, since the added value of the other tests is not known. One or more of the above tests may be used in addition to, or instead of, CT scan. In addition to the above tests, diagnostic laparoscopy and diagnostic ultrasound are other tests that may be used in the differential diagnosis of focal pancreatic lesions. Laparoscopy may avoid the problem with tumour seeding associated with percutaneous biopsy. However, these tests are not used routinely. The accuracy of diagnostic laparoscopy and diagnostic ultrasound are being assessed in a different review (Gurusamy 2011a). Serum carbohydrate antigen 19‐9 (CA 19‐9) is a substance released into the bloodstream by both cancer cells and normal cells. Too much CA 19‐9 in the blood can be a sign of pancreatic cancer or other types of cancer or conditions. The amount of CA 19‐9 in the blood can be used to help keep track of how well cancer treatments are working or if cancer has come back. It is a type of tumour marker (National Cancer Institute 2011f) and may be used in conjunction with another imaging modality in the assessment of the focal pancreatic lesion. This will be considered in a different review (Gurusamy 2011b). The diagnostic accuracy of the index tests considered in this review will be compared with these alternative tests.
Rationale
Different imaging modalities use different methods to differentiate normal and diseased tissues. Endoscopic ultrasound is closer to the tissues and hence can use high‐frequency ultrasound waves, which have better resolution but poorer penetration than an external ultrasound. Image‐guided biopsy can be performed and the tissue can be examined under the microscope or proteomic profiling can be performed to differentiate between the different types of focal pancreatic lesion.
Accurate characterisation of lesions will help in the management of the patients. Patients with cancerous lesions will be offered surgery if there is no distant spread of cancer and if they are anaesthetically fit for major surgery. Patients with cancerous lesions who are not eligible for surgery because of distant spread of cancer or because of anaesthetic fitness for major surgery will be offered other treatments such as chemotherapy. Patients with precancerous lesions may also undergo surgery depending upon the clinicians' and patients' preferences. Unnecessary major surgery can be avoided in patients with benign lesions.
Currently, there is no Cochrane review of studies assessing the diagnostic accuracy of different imaging modalities in the assessment of focal pancreatic lesions.
Objectives
To determine and compare the diagnostic accuracy of various imaging modalities in detecting cancerous and precancerous lesions in patients with focal pancreatic lesions.
Investigation of sources of heterogeneity
We will explore the following sources of heterogeneity.
-
Studies of low risk of bias versus those of unclear or high risk of bias (as assessed by the revised QUADAS tool as recommended by the Cochrane Screening and Diagnostic Tests Methods Group) (Whiting 2006). In particular, we consider the studies which have been classified as 'yes' in the items differential verification, uninterpretable results and withdrawals as the most important sources of heterogeneity.
-
Full text publications versus abstracts (this might give a clue on the publication bias since there may be an association between the results of the study and the study reaching full publication (Eloubeidi 2001).
-
Prospective studies versus retrospective studies.
-
Different types of reference standard.
-
Symptomatic versus asymptomatic lesions (the presence of symptoms may increase the pretest probability).
-
Solid versus cystic lesions (as the diagnostic accuracy of the imaging modalities may vary depending upon whether the lesion is solid or cystic).
-
Patients with chronic pancreatitis versus those without chronic pancreatitis.
-
Different criteria used by authors to classify the lesions.
-
Single imaging versus sequential imaging (repeated imaging).
-
Different intervals of sequential imaging (e.g. imaging every six months versus annual review).
Methods
Criteria for considering studies for this review
Types of studies
We will include studies reporting on cross‐sectional information of the index test and reference test in the appropriate patient population (see below) irrespective of language or publication status or whether the data were collected prospectively or retrospectively. However, we will exclude case series in which only true positive results or true negative results are reported without any information on the other patients who underwent the test.
Participants
Patients with focal pancreatic lesions.
Index tests
CT scan, MRI scan, PET scan, EUS, EUS elastography, and EUS‐guided biopsy either alone or in combination as replacement for major surgery for diagnostic purposes.
We will accept the criteria stated by the authors to classify the lesion as benign, precancerous, and cancerous for different imaging modalities.
Comparator tests
There is no standard algorithm in the diagnosis or management of focal pancreatic lesions. Other tests that may be used in the diagnosis of focal pancreatic lesions include diagnostic laparoscopy, laparoscopic ultrasound, serum levels of CA 19‐9, and surgical resection (surgical resection may be considered diagnostic when the diagnosis is uncertain after all other diagnostic modalities have been attempted).
Target conditions
Benign versus precancerous and cancerous lesions (including the type of cancerous lesion).
Reference standards
We will accept the following reference standards.
-
Histopathological examination of the entire lesion by surgical resection (gold standard). This will classify the lesion as benign, precancerous, or cancerous.
-
Histopathological examination (irrespective of how the tissues were obtained for histopathological examination) in patients with positive test (for cancerous or precancerous lesions) and clinical follow‐up by a doctor (with or without sequential follow‐up with imaging) of all patients with negative test for a period of at least six months and for a maximum period of 24 months. Until a definitive diagnosis is available, biopsy is generally avoided because of the fear of seeding of cancer cells. So, we anticipate that the tissues obtained for histopathological examination are obtained from surgical resection. It is unlikely that patients who have low likelihood for cancer based on the clinical symptoms and signs, and results of tests (may include the results of index test) are subject to surgery or biopsy. Even if a biopsy is performed in such patients, a cancerous lesion or precancerous lesion cannot be ruled out because of sampling error. So, such patients are usually followed up clinically with sequential imaging. Most types of pancreatic cancers will cause clinical deterioration or increase in tumour size during a period of six months and so we will accept clinical follow‐up or sequential follow‐up imaging (irrespective of the modality of the imaging) of all patients with a negative biopsy or no biopsy for at least six months as one of the reference standards. The choice of a maximum period of 24 months was based on an arbitrary choice based on the low probability of precancerous lesions becoming cancerous during 24 months.
Search methods for identification of studies
Electronic searches
We will search the following databases.
-
The Cochrane Register of Diagnostic Test Accuracy Studies (latest issue) (Appendix 1).
-
The Cochrane Central Register of Controlled Trials (CENTRAL) in The Cochrane Library (latest issue) (Appendix 1).
-
MEDLINE via PubMed (January 1946 to date of search) (Appendix 2).
-
EMBASE via OvidSP (January 1947 to date of search) (Appendix 3).
-
Science Citation Index Expanded via ISI Web of Knowledge (January 1980 to date of search) (Appendix 4).
Searching other resources
We will search the references of included studies to identify further studies (Horsley 2011). We will also search for additional articles related to included studies by performing the 'related search' function in MEDLINE (PubMed) and EMBASE (OvidSP) and 'citing reference' search (search the articles that cited the included articles) (Sampson 2008) in Science Citation Index Expanded and EMBASE (OvidSP).
Data collection and analysis
Selection of studies
Two authors will search the references independently for identification of relevant studies. We will obtain full texts for the references that at least one of the authors consider relevant. We will use the obtained full texts to further exclude irrelevant references. References to studies that meet the inclusion criteria will be selected for data extraction. Any differences in study selection will be arbitrated by BR Davidson.
Data extraction and management
Two review authors will extract the following data from each included study independently.
-
First author of report.
-
Year of publication of report.
-
Study design (prospective or retrospective; cross‐sectional studies or randomised clinical trials).
-
Inclusion and exclusion criteria for individual studies.
-
Total number of patients.
-
Number of females.
-
Mean age of the participants.
-
Criteria used for classification of lesions.
-
Preoperative tests carried out prior to index test.
-
Index test.
-
Reference standard.
-
True positive (TP), false positive (FP), true negative (TN), and false negative (FN) data.
Main analysis
The unit of analysis will be the patient. If available, we will extract the TP, FP, TN, and FN information for the following situations.
-
Benign lesions with no malignant potential (negative test) versus precancerous or cancerous lesions (positive test) (this will help in the determination of whether the patient needs further follow‐up).
-
Cancerous lesions (positive test) versus non‐cancerous lesions (negative test) (this will help in the determination of whether the patient needs immediate surgery).
-
In the group of patients with precancerous or cancerous lesions(i.e. those with positive test in the analysis of benign lesions with no malignant potential (negative test) versus precancerous or cancerous lesions (positive test)), we will extract the TP, FP, TN, and FN information for precancerous lesions (negative test) versus cancerous lesions (positive test) (this will help in the assessment of pros and cons of surgery ‐ surgery is the only curative option for cancerous lesions; follow‐up may be an option in precancerous lesions).
We will extract the information on indeterminate results separately from the TP, FP, TN, and FN data. There is no standard algorithm of management of patients with indeterminate results in the first scan. Some surgeons may recommend surgical resection for indeterminate lesions. Other surgeons may recommend additional scans or sequential follow‐up imaging. So, a sensitivity analysis will be performed as mentioned in the section on 'Sensitivity analysis'.
For tests performed for sequential follow‐up imaging (repeated testing strategy), the TP, FP, TN, and FN data will be obtained for the strategy as a whole. Increase in size or change in the lesion on sequential follow‐up imaging (performed within 12 months) will be considered as positive index test. If the lesion remains static (or becomes smaller) in size without any change in the characteristics of the lesion, this will be considered as negative index test. Majority of the surgeons will recommend further follow‐up imaging or no follow‐up if the sequential follow‐up image shows no change in the lesion and there is no clinical deterioration for the comparison between precancerous and cancerous lesions. So, indeterminate results on sequential follow‐up imaging will be considered as negative results for this comparison.
Further information will be sought from the authors of the studies if necessary. Any differences between the review authors will be resolved by discussion.
Assessment of methodological quality
Two review authors will assess the quality of the studies independently using the QUADAS‐2 assessment tool (quality assessment of studies of diagnostic accuracy included in systematic reviews) (Whiting 2006; Whiting 2011). Any differences in the methodological quality assessment will be resolved by BR Davidson. Further information will be sought from the authors of the studies to assess the methodological quality of the studies accurately.
The quality items derived from the QUADAS‐2 tool that will be assessed and the way that they will be interpreted are stated in Table 1.
| Domain 1: patient sampling | Signalling question | Signalling question | Signalling question | Risk of bias | Concerns for applicability |
| Patient sampling | Was a consecutive or random sample of patients enrolled? | Was a case‐control design avoided? | Did the study avoid inappropriate exclusions? | Could the selection of patients have introduced bias? | Are there concerns that the included patients and setting do not match the review question? |
| Yes: all consecutive patients or random sample of patients with focal pancreatic lesions were enrolled No: selected patients were enrolled Unclear: this was not clear from the report | Yes: case‐control design was avoided No: case‐control design was not avoided Unclear: this was not clear from the report | Yes: the study avoided inappropriate exclusions (i.e. difficult to diagnose patients) No: the study excluded patients inappropriately Unclear: this was not clear from the report | Low risk: 'yes' for all signalling questions High risk: 'no' or 'unclear' for at least 1 signalling question | Low concern: the selected patients represent the patients in whom the tests will be used in clinical practice (please see diagnostic pathway (Figure 1)) High concern: there is high concern that patient selection was performed in such a way that the included patients did not represent the patients in whom the tests will be used in clinical practice | |
| Domain 2: index test | |||||
| Index test(s) | Were the index test results interpreted without knowledge of the results of the reference standard? | If a threshold was used, was it pre‐specified? | Could the conduct or interpretation of the index test have introduced bias? | Are there concerns that the index test, its conduct, or interpretation differ from the review question? | |
| Yes: index test results were interpreted without knowledge of the results of the reference standard No: index test results were interpreted with knowledge of the results of the reference standard Unclear: this was not clear from the report | Yes: if the criteria for a positive test were pre‐specified No: if the criteria for a positive test were not pre‐specified Unclear: this was not clear from the report | Low risk: 'yes' for all signalling questions High risk: 'no' or 'unclear' for at least 1 of the 2 signalling questions | High concern: there is high concern that the conduct or interpretation of the index test differs from the way it is likely to be used in clinical practice Low concern: there is low concern that the conduct or interpretation of the index test differs from the way it is likely to be used in clinical practice | ||
| Domain 3: reference standard | |||||
| Target condition and reference standard(s) | Is the reference standard likely to classify the target condition correctly? | Were the reference standard results interpreted without knowledge of the results of the index tests? | Could the reference standard, its conduct, or its interpretation have introduced bias? | Are there concerns that the target condition as defined by the reference standard does not match the review question? | |
| Yes: histopathological examination of the entire lesion by surgical resection No: histopathological examination (irrespective of how the tissues were obtained for histopathological examination) in patients with positive test (for cancerous or precancerous lesions) and clinical follow‐up by a doctor (with or without sequential follow‐up with imaging) of all patients with negative test for a period of at least 6 months and for a maximum period of 24 months Unclear: this was not clear from the report. Such studies will be excluded Yes: reference standard results were interpreted without knowledge of the results of the index test No: reference standard results were interpreted with the knowledge of the results of the index test Unclear: this was not clear from the report | Low risk: 'yes' for all signalling questions High risk: 'no' or 'unclear' for at least 1 of the 2 signalling questions | Low concern: histopathological examination of the entire lesion by surgical resection High concern: histopathological examination (irrespective of how the tissues were obtained for histopathological examination) in patients with positive test (for cancerous or precancerous lesions) and clinical follow‐up by a doctor (with or without sequential follow‐up with imaging) of all patients with negative test for a period of at least 6 months and for a maximum period of 24 months | |||
| Domain 4: flow and timing | |||||
| Flow and timing | Was there an appropriate interval between index test and reference standard? | Did all patients receive the same reference standard? | Were all patients included in the analysis? | Could the patient flow have introduced bias? | |
| Yes: histopathological examination of the entire lesion (gold standard) ‐ performed within 2 months (chosen arbitrarily). Histopathological examination (irrespective of how the tissues were obtained for histopathological examination) in patients with positive test (for cancerous or precancerous lesions) performed within 2 months and clinical follow‐up (including sequential follow‐up with imaging) of all patients with negative test for a period of at least 6 months. No: the histopathological examination was performed beyond 2 months of the index tests. The clinical follow‐up (including sequential follow‐up imaging) was performed less than 6 months of the index test. This is because some tumours may be slow‐growing Unclear: this was not clear from the report | Yes: histopathological examination of the entire lesion by surgical resection No: histopathological examination (irrespective of how the tissues were obtained for histopathological examination) in patients with positive test (for cancerous or precancerous lesions) and clinical follow‐up by a doctor (with or without sequential follow‐up with imaging) of all patients with negative test for a period of at least 6 months and for a maximum period of 24 months Unclear: this was not clear from the report. Such studies will be excluded | Yes: all patients meeting the selection criteria (selected patients) were included in the analysis, or data on all the selected patients were available so that a 2 x 2 table including all selected patients could be constructed No: not all patients meeting the selection criteria were included in the analysis or the 2 x 2 table could not be constructed using data on all selected patients Unclear: this was not clear from the report | Low risk: 'yes' for all signalling questions High risk: 'no' or 'unclear' for at least 1 signalling question |
Statistical analysis and data synthesis
The data obtained from the various studies will be presented in the form of forest plots of sensitivity and specificity with their corresponding 95% confidence intervals. We will assess the correlation between the sensitivity and specificity obtained from the different studies by calculating the Spearman correlation coefficient in order to assess whether there is any intrinsic threshold effect due to the different criteria for classifying the lesions used by different authors. The data will be combined using the hierarchical summary receiver operating characteristics (HSROC) model (Rutter 2001) using the METADAS macro developed by the Systematic Reviews of Diagnostic Test Accuracy (SRDTA) Working Group (Takwoingi 2008) in the SAS 9.2 statistical software (SAS Institute Inc., Cary, NC, USA) for those tests with an explicit cut‐off threshold. This allows inclusion of studies using any cut‐off threshold. This will give us an idea of whether one test is generally better than the other. We will use the bivariate model (Reitsma 2005) to calculate the summary sensitivity and specific at specific thresholds for these tests using the same METADAS macro. We will also use the bivariate model for those tests that do not have an explicit cut‐off threshold. METADAS allows meta‐analysis of diagnostic test accuracy studies in which both the index test under study and the reference test (gold standard) are dichotomous. It involves statistical distributions at two levels. At the lower level, it models the cell counts in the 2 x 2 tables extracted from each study using binomial distributions and logistic (log‐odds) transformations of proportions, At the higher level, random study effects are assumed to account for heterogeneity in diagnostic test accuracy between studies beyond that accounted for by sampling variability at the lower level (Macaskill 2009). We will first assess the diagnostic accuracy of the index test in distinguishing benign lesions with no malignant potential from precancerous or cancerous lesions (first main analysis). We will also assess the diagnostic accuracy of the tests in distinguishing cancerous lesions versus non‐cancerous lesions (second main analysis). In the first main analysis, we will assess the diagnostic accuracy of the tests in distinguishing precancerous lesions from cancerous lesions (additional analysis) in patients who have precancerous or cancerous lesions. We will also create a graph of pretest probabilities (from 0.05 to 1.00) versus corresponding post‐test probabilities following a positive test and for a negative test based on the summary sensitivity and specificity.
Investigations of heterogeneity
We will visually inspect forest plots of sensitivity and specificity and the receiver operating characteristics (ROC) curve to identify heterogeneity. We will explore heterogeneity by using the different sources of heterogeneity as covariates in the METADAS macro. We will assess whether there is a statistically significant difference in the likelihood ratios to identify heterogeneity.
Sensitivity analyses
In the presence of indeterminate results (due to any reason) for the initial test, we will consider two scenarios. We will consider the patients with indeterminate results as positive for the test as some surgeons will recommend surgical resection for indeterminate lesions. We will also consider the indeterminate results as negative for the test, since some surgeons will recommend sequential follow‐up imaging. We will assess the diagnostic accuracy in both these scenarios.
We will also assess the comparative performance of tests by direct comparison (i.e. the tests performed in the same patient) versus indirect comparison (the tests performed in different patients across studies).
Clinical pathway.
Ca 19‐9: carbohydrate antigen 19‐9;
CT: computerised tomogram;
EUS: endoscopic ultrasound;
MRI: magnetic resonance imaging;
PET: positron emission tomogram.
| Domain 1: patient sampling | Signalling question | Signalling question | Signalling question | Risk of bias | Concerns for applicability |
| Patient sampling | Was a consecutive or random sample of patients enrolled? | Was a case‐control design avoided? | Did the study avoid inappropriate exclusions? | Could the selection of patients have introduced bias? | Are there concerns that the included patients and setting do not match the review question? |
| Yes: all consecutive patients or random sample of patients with focal pancreatic lesions were enrolled No: selected patients were enrolled Unclear: this was not clear from the report | Yes: case‐control design was avoided No: case‐control design was not avoided Unclear: this was not clear from the report | Yes: the study avoided inappropriate exclusions (i.e. difficult to diagnose patients) No: the study excluded patients inappropriately Unclear: this was not clear from the report | Low risk: 'yes' for all signalling questions High risk: 'no' or 'unclear' for at least 1 signalling question | Low concern: the selected patients represent the patients in whom the tests will be used in clinical practice (please see diagnostic pathway (Figure 1)) High concern: there is high concern that patient selection was performed in such a way that the included patients did not represent the patients in whom the tests will be used in clinical practice | |
| Domain 2: index test | |||||
| Index test(s) | Were the index test results interpreted without knowledge of the results of the reference standard? | If a threshold was used, was it pre‐specified? | Could the conduct or interpretation of the index test have introduced bias? | Are there concerns that the index test, its conduct, or interpretation differ from the review question? | |
| Yes: index test results were interpreted without knowledge of the results of the reference standard No: index test results were interpreted with knowledge of the results of the reference standard Unclear: this was not clear from the report | Yes: if the criteria for a positive test were pre‐specified No: if the criteria for a positive test were not pre‐specified Unclear: this was not clear from the report | Low risk: 'yes' for all signalling questions High risk: 'no' or 'unclear' for at least 1 of the 2 signalling questions | High concern: there is high concern that the conduct or interpretation of the index test differs from the way it is likely to be used in clinical practice Low concern: there is low concern that the conduct or interpretation of the index test differs from the way it is likely to be used in clinical practice | ||
| Domain 3: reference standard | |||||
| Target condition and reference standard(s) | Is the reference standard likely to classify the target condition correctly? | Were the reference standard results interpreted without knowledge of the results of the index tests? | Could the reference standard, its conduct, or its interpretation have introduced bias? | Are there concerns that the target condition as defined by the reference standard does not match the review question? | |
| Yes: histopathological examination of the entire lesion by surgical resection No: histopathological examination (irrespective of how the tissues were obtained for histopathological examination) in patients with positive test (for cancerous or precancerous lesions) and clinical follow‐up by a doctor (with or without sequential follow‐up with imaging) of all patients with negative test for a period of at least 6 months and for a maximum period of 24 months Unclear: this was not clear from the report. Such studies will be excluded Yes: reference standard results were interpreted without knowledge of the results of the index test No: reference standard results were interpreted with the knowledge of the results of the index test Unclear: this was not clear from the report | Low risk: 'yes' for all signalling questions High risk: 'no' or 'unclear' for at least 1 of the 2 signalling questions | Low concern: histopathological examination of the entire lesion by surgical resection High concern: histopathological examination (irrespective of how the tissues were obtained for histopathological examination) in patients with positive test (for cancerous or precancerous lesions) and clinical follow‐up by a doctor (with or without sequential follow‐up with imaging) of all patients with negative test for a period of at least 6 months and for a maximum period of 24 months | |||
| Domain 4: flow and timing | |||||
| Flow and timing | Was there an appropriate interval between index test and reference standard? | Did all patients receive the same reference standard? | Were all patients included in the analysis? | Could the patient flow have introduced bias? | |
| Yes: histopathological examination of the entire lesion (gold standard) ‐ performed within 2 months (chosen arbitrarily). Histopathological examination (irrespective of how the tissues were obtained for histopathological examination) in patients with positive test (for cancerous or precancerous lesions) performed within 2 months and clinical follow‐up (including sequential follow‐up with imaging) of all patients with negative test for a period of at least 6 months. No: the histopathological examination was performed beyond 2 months of the index tests. The clinical follow‐up (including sequential follow‐up imaging) was performed less than 6 months of the index test. This is because some tumours may be slow‐growing Unclear: this was not clear from the report | Yes: histopathological examination of the entire lesion by surgical resection No: histopathological examination (irrespective of how the tissues were obtained for histopathological examination) in patients with positive test (for cancerous or precancerous lesions) and clinical follow‐up by a doctor (with or without sequential follow‐up with imaging) of all patients with negative test for a period of at least 6 months and for a maximum period of 24 months Unclear: this was not clear from the report. Such studies will be excluded | Yes: all patients meeting the selection criteria (selected patients) were included in the analysis, or data on all the selected patients were available so that a 2 x 2 table including all selected patients could be constructed No: not all patients meeting the selection criteria were included in the analysis or the 2 x 2 table could not be constructed using data on all selected patients Unclear: this was not clear from the report | Low risk: 'yes' for all signalling questions High risk: 'no' or 'unclear' for at least 1 signalling question |
