SEARCH

SEARCH BY CITATION

Keywords:

  • Diagnosis/analysis;
  • diagnosis/diagnostic use;
  • diagnosis/methods;
  • epidemiology/diagnosis;
  • genes;
  • KRAS oncogene;
  • oncogene protein p21 ras;
  • pancreatic neoplasms/diagnosis;
  • ras;
  • research/methods

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure of interests
  9. Address
  10. References
  11. Supporting Information

Eur J Clin Invest 2011; 41 (7): 793–805

Abstract

Background  More clinically meaningful diagnostic tests are needed in exocrine pancreatic cancer (EPC). K-ras mutations are the most frequently acquired genetic alteration in EPC. We analysed the diagnostic utility of detecting K-ras mutations through a systematic analysis of the literature.

Methods  We searched PubMed using suitable medical subject headings and text words. Original research articles that evaluated the diagnostic accuracy of detecting K-ras mutations for diagnosis of EPC were selected. Two investigators independently extracted data from each study regarding the methodology used, the methodological quality of the study, the diagnostic accuracy reported and the authors’ conclusions about clinical applicability of the test. Combined estimates for the sensitivity and specificity of K-ras were determined using bivariate meta-analysis; heterogeneity was explored using meta-regression.

Results  We assessed 34 studies from 30 published articles. The research reports were prone to numerous methodological biases and often lacked vital information for assessing external validity. The sensitivity of detecting K-ras status ranged from 0% through 100%, and the specificity from 58% through 100%. Diagnostic accuracy was highest when cytohistological samples were used: sensitivity and specificity were 76·5% (66·7–84·2) and 91·8% (87·6–94·1), respectively. Studies conducted in a clinically relevant population observed lower accuracy than case–control designs (68·4% vs. 82·7%).

Conclusions  Because of the numerous methodological limitations of studies, the utility of analysing K-ras mutations for the diagnosis of EPC remains unknown. Flaws in diagnostic biomarkers with well-established biological properties, as K-ras, become even more relevant when the promises of ‘personalized medicine’ are pondered.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure of interests
  9. Address
  10. References
  11. Supporting Information

There were an estimated 58,100 new cases of pancreatic cancer in the European Union in 2006, and some 64 000 deaths caused by the disease [1], although differences in mortality rates across countries and over time may partly be because of changing diagnostic and death certification procedures [2]. At the time of diagnosis, patients with pancreatic cancer usually have advanced disease and treatment with curative intent is rare [3]; thus, the mortality rate remains similar to the incidence rate. Diagnosis of exocrine pancreatic cancer (EPC) is particularly challenging because early symptoms and laboratory findings are usually nonspecific [4], and imaging tests are often insufficient for precise staging, assessment of resectability or accurate discrimination between malignant and benign disease [5–7]. CA 19-9 has been considered the best-validated serum marker for patients with EPC. CA 19-9 has a somewhat limited use in the early diagnosis of EPC because of its poor sensitivity and specificity, and it is not especially useful in monitoring the progression of the disease; however, CA 19-9 may complement other diagnostic procedures and assist in predicting survival [8]. In any case, more clinically meaningful and efficient diagnostic strategies – including molecular markers – are needed for EPC.

Present in 75–90% of tumours, activating point mutations in the K-ras oncogene are the most frequently acquired genetic alteration in EPC [9–11]. K-ras mutations can be detected – with substantially different accuracy – in pancreatic tissue [6,8,9], pancreatic juice [12], bile [13], plasma [14] and stools [15]. K-ras tests may hence have different diagnostic properties. Mutations in the K-ras gene have also been detected in the normal pancreas and in patients with benign pancreatic diseases [16,17], a fact that may limit the clinical utility of K-ras tests for the diagnosis of EPC. Numerous studies have been carried out to determine the diagnostic utility of detecting K-ras mutations, but reports are heterogeneous and conflicting. On one hand, the basis for variation in the studies may be linked to study size (random error) and to differences in the type of biological sample used for K-ras determination. On the other hand, methodological characteristics of the studies – notably, the strategy to recruit patients – may have biased estimates of diagnostic accuracy (systematic error) [3,18–20].

A research structure for the development and evaluation of new diagnostic tests has long been defined [21–25]. Unfortunately, it is rarely applied comprehensively [7,26–28]; as a result, diagnostic research tends to be of lower quality than therapeutic research [26,29,30]. Before a diagnostic test is introduced into clinical practice, it is imperative that it be tested in a series of patients similar to those who would receive the test in practice [18–26,31–33]. Yet, some studies select a group with an established diagnosis and compare it with a control group whose members clearly do not have the disease of interest; this design feature is of low clinical relevance and usually leads to inflated estimates of diagnostic accuracy [18–23,34]. Furthermore, given the deep anatomical situation of the pancreas, and the resulting difficulty in obtaining cytohistological samples, diagnostic studies of EPC may be especially susceptible to selection biases [3,6,35–39].

Methodological flaws well known in traditional diagnostic research have become more challenging with some of the new molecular technologies. Thus, opportunities for error and variation could hamper the rapid development of individualized medicine [40].

The objective of the present review was to synthesize and analyse the results from a systematic selection of research papers that evaluated the utility of detecting the K-ras mutational status for the diagnosis of EPC. We sought to describe the methodological characteristics of these articles and to evaluate how such features affected the diagnostic accuracy observed. Furthermore, we aimed to describe the authors’ conclusions regarding the potential clinical usefulness of K-ras tests for the diagnosis EPC, either as an independent diagnostic tool or in combination with other existing diagnostic procedures.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure of interests
  9. Address
  10. References
  11. Supporting Information

Search and selection of articles

We identified studies by carrying out a systematic search of the PubMed database using the medical subject headings (MeSH) ‘Pancreatic neoplasms’, ‘Diagnosis’, ‘Genes, ras’ and ‘Sensitivity and Specificity’, as well as various text words such as ‘pancreas’, ‘pancreatic’, ‘cancer’, ‘carcinoma’, ‘tumour’, ‘predictive value’, ‘accuracy’, ‘screening’, ‘KRAS’, ‘K-ras’kirsten-ras’ and ‘ki-ras’. We also performed a supplementary search in the EMBASE database and scanned the reference sections of key articles. A detailed report of the search terms used can be found in Table S1. The most recent search was carried out on 16th July 2010. The titles and abstracts of all potential articles were reviewed, and articles were preselected based on the following criteria: original research articles that evaluated the diagnostic accuracy of detecting K-ras mutations for diagnosis of EPC in at least 50 patients. Studies that evaluated the diagnostic validity of K-ras in combination with other diagnostic procedures were included, and there was no limit to the type of patient sample used for DNA extraction and K-ras determination (i.e. tissue, blood, pancreatic juice,) On reviewing the full text of preselected articles, studies that used the term ‘pancreatic cancer’ generically and did not specify the histology were excluded (e.g. if we were unable to assess whether they included endocrine tumours). Articles that evaluated the sensitivity of K-ras alone and did not include a comparison group were also excluded. A flow diagram describing the search and selection process can be found in Fig. 1.

image

Figure 1.  Literature search and selection process.*Exact search terms can be found in the supporting information, supplementary Table S1.

Download figure to PowerPoint

Data extraction

Using a structured form, two investigators (LAP, BL) independently extracted the following data from the articles: Author, year of publication, journal, number of patients initially included in the study, number of patients that contributed to the evaluation of K-ras and whether the study evaluated the diagnostic accuracy of another marker in addition to K-ras. We also recorded the patient recruitment strategy using the following categories: (i) case–control; (ii) consecutive patients with suspected EPC; (iii) nonconsecutive patients with suspected EPC; (iv) consecutive patients with suspected biliary or pancreatic disease; (v) nonconsecutive patients with suspected biliary or pancreatic disease; (vi) consecutive patients undergoing a certain procedure; (vii) nonconsecutive patients undergoing a certain procedure and (viii) unclear. We evaluated the quality of the studies using adapted quality criteria from the QUADAS guideline [41], as well as some additional items specific for EPC. We recorded the type of sample used for K-ras analysis, how the material was obtained and if this procedure was invasive (blood extraction was considered not invasive, fine needle aspiration was considered minimally invasive, endoscopic retrograde cholangio-pancreatography moderately invasive and surgery very invasive).

To evaluate the authors’ conclusions regarding the clinical application of detecting K-ras mutations for the diagnosis of EPC, two researchers (LAP, BL) independently recorded all text in the abstract and/or discussion of each study that referred to the clinical application of the test (i.e. we extracted all sentences that dealt with the clinical application of K-ras mutations). Each extracted text was then coded as definitively favourable, moderately favourable or not favourable depending on the type of language used and then as a function of the diagnostic context to which it referred (i.e. independent use, as an adjunct to another procedure, in limited situations, or not at all). This information was then used to generate the following classification: (i) not favourable in any context; (ii) moderately or definitively favourable to application only in a limited context (i.e. certain subgroups of patients); (iii) moderately or definitively favourable towards application as an adjunct to another diagnostic procedure; and (iv) moderately or definitively favourable to application as an independent diagnostic procedure. When studies evaluated K-ras mutations in more than one type of sample or for more than one patient subgroup, the texts referring to the clinical application of each setting were extracted, and then the article as a whole was classified as above. In practice, the distinction between subgroups was not always clear-cut, and authors’ conclusions tended to refer to the most promising sample type or patient group only.

Finally, we extracted the raw data for the analysis of sensitivity and specificity (that is, the number of true positive, false negative, true negative and false positive results) for the comparison of patients diagnosed with EPC vs. all other patients included. Furthermore, when available, we recorded the information regarding the diagnostic accuracy of K-ras in combination with another procedure, and when studies suggested that detecting K-ras was only useful in a limited context, we recorded what that context was and the diagnostic accuracy that they reported in this context.

When the articles described two different types of sample or presented two separate patient populations, the results were recorded as two separate studies. Overall, the agreement between the two reviewers in the data extraction was 85·8%. Discrepancies were resolved by consensus.

Data analysis

Normality of continuous variables was assessed visually and numerically with the Shapiro–Wilk test. Study size and disease prevalence were described using the median and interquartile range, according to relevant subgroups. We summarized the diagnostic accuracy of the studies using the ‘metandi’ module for Stata [42,43] and bivariate meta-analysis of joint pairs of sensitivity and specificity. Meta-regression was performed as described by Reitsma et al. [44] in a subset of studies, in an attempt to explain the observed heterogeneity and to explore how certain methodological characteristics influenced the observed sensitivity and specificity. Comparisons between subgroups were made using this method. Descriptive analyses were carried out using Stata/SE 8·0, Fig. 1 was created with Intercooled Stata 9·0, and bivariate meta-regression was performed using SAS 9·1·3 (SAS Institute, Cary, NC, USA).

Reporting

Reporting of the study conforms to the PRISMA statement for reporting systematic reviews and meta-analyses. [45,46]

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure of interests
  9. Address
  10. References
  11. Supporting Information

We identified 30 articles published from 1991 through 2010 (20 years). The full references can be found in the Annex; the characteristics of each individual study (stated objectives, authors’ conclusions and other variables, such as sample size, patient selection strategy, type of sample, procedure for obtaining the sample and the diagnostic accuracy calculated) can be found in the Tables S2–S4. Twenty-seven of the 30 studies (90·0%) detected K-ras mutations in codon 12, and over half (17 studies, 56·7%) used restriction fragment length polymorphism (RFLP) PCR. Only seven studies (23·3%) reported the analytical sensitivity of the technique. A table summarizing the analytical details of the K-ras mutational testing can be found in the Table S5.

Four studies presented the results for more than one type of sample (for example, pancreatic juice and stools) or for more than one patient subgroup (for example, according to the presence or absence of a pancreatic mass). Therefore, of 30 studies, we present the results from a total of 34 independent patient series, herein referred to as individual studies.

Patients included in each study

The number of patients described as having entered the study ranged widely from 51 through 532. The median size was 78·5 (interquartile range [IQR]: 62–109). Thirteen studies (38·2%) did not report the K-ras mutational status for all patients initially recruited for the study. These studies determined K-ras status in a median of 93·1% of the included patients (range: 53·2–95·2%). One study reported the results for K-ras analysis in more patients than were described as having entered the study [45]. The number of patients used to estimate the diagnostic accuracy of K-ras in the studies ranged from 51 through 358, with a median of 77·5 (IQR: 57–105).

Nineteen studies (55·9%) used a diagnostic case–control design, and three (8·8%) did not clearly describe how they had selected patients. The remaining 12 studies (35·3%) used a clinically relevant patient series, of which only 4 (11·8%) had a consecutive series of patients with suspected EPC (Table 1). In 10 studies (29·4%), participants were recruited in a consecutive manner. The prevalence of EPC also ranged broadly from 9·3% through 81·8%. A graphical representation of the study size and the proportion of patients with a final diagnosis of EPC is shown in Fig. 2. This graph does not include the studies which used a case–control design because the prevalence of disease is artificial in these populations. While in case–control designs the median prevalence of EPC was 43·9%, in studies that included a more clinically relevant patient series, the corresponding figure tended to be even higher. For instance, the median prevalence of EPC in patients recruited because of suspected EPC was 66·3%. The prevalence was similar in studies that recruited patients based on a clinical suspicion of unspecified biliary or pancreatic disease (57·9%), or patients recruited because they were undergoing a certain procedure at the time of the study (62·4%). Nine studies (26·5) reported the tumour stage of the cancer patients included. On average, such studies included 45·3% (SD: 24·5%) stage IV tumours.

Table 1.   Patient selection, type of sample and procedures in 34 studies evaluating the diagnostic utility of detecting K-ras mutations in the diagnosis of exocrine pancreatic cancer in at least 50 patients
 N (%)
Patient selection design
 Case–control19 (55·9)
 Suspicion of exocrine pancreatic cancer4 (11·8)
 Suspicion of biliary or pancreatic disease3 (8·8)
 Patients undergoing a certain procedure5 (14·7)
 Unclear3 (8·8)
Patient selection strategy
 Consecutive10 (29·4)
 Nonconsecutive24 (70·6)
Type of sample
 Tissue17 (50·0)
 Pancreatic juice, bile or other digestive secretions9 (26·5)
 Serum or plasma4 (11·8)
 Stool3 (8·8)
 Various1 (2·9)
Invasiveness of procedure to obtain sample
 Noninvasive7 (20·6)
 Minimally invasive10 (29·4)
 Moderately invasive9 (26·5)
 Highly invasive8 (23·5)
image

Figure 2.  Study size and disease prevalence according to patient recruitment in 15 studies evaluating the diagnostic validity of K-ras mutations in the diagnosis of exocrine pancreatic cancer (EPC).

Download figure to PowerPoint

Type of sample and procedures

The K-ras mutational status was determined in tissue in 17 studies (50·0%), in pancreatic juice, bile or other secretions of the digestive tract in nine studies (26·5%) and in serum or plasma in four studies (11·8%), while three studies (8·8%) used stool samples and 1 (2·9%) used more than one type of sample and did not separate the results (Table 1). The method for obtaining the sample for K-ras analysis was considered to be highly invasive in eight studies (23·5%). In the studies that detected K-ras in tissue, we sought to determine whether they used histological or cytological samples. Six studies (35·3%) used histological samples, eight (47·1%) used cytological specimens, and in three studies, the type of sample was not clear or a mixture of both cytological and histological samples were used.

Quality analysis

We included 14 items to evaluate the quality of the studies; a table reflecting how the individual studies scored for each quality item can be found in the Table S6. Furthermore, a summary of the quality of the 34 studies can be found in Table 2. Notably, the selection criteria were not clearly described in 22 studies (64·7%), and various studies did not report socio-demographic or clinical details of the final patient population (nine studies, 26·5%, and 16 studies, 47·1%, respectively). Under half of the studies stated that the definitive diagnosis of the patients was made without knowledge of the results of K-ras analysis (13 studies, 38·2%), while even fewer studies (9, 26·5%) stated that the K-ras mutational status of the patients was interpreted blinded to the definitive diagnosis. Both situations may lead to reviewer bias and inflated estimates of diagnostic accuracy. Of 15 studies that did not report K-ras results for all patients who entered the study, five (33·3%) failed to explain why this had occurred. Of the 10 that did explain, common reasons were failures of DNA amplification and problems with sample collection.

Table 2.   Quality analysis of 34 studies evaluating the diagnostic utility of detecting K-ras mutations in the diagnosis of exocrine pancreatic cancer in at least 50 patients
 Yes (%)No (%)Unclear (%)
Q1. Was the spectrum of patients representative of the patients who will receive  the test in practice?6 (17·6)27 (79·4)1 (2·9)
Q2. Were selection criteria clearly described?12 (35·3)22 (64·7)0 (0·0)
Q3. Were the socio-demographic details of the patients reported (age, sex, etc.)?23 (67·6)9 (26·5)2 (5·9)
Q4. Were relevant clinical details of the patients reported (cancer stage, co-morbidities, etc.)?18 (52·9)16 (47·1)0 (0·0)
Q5. Were all of the cases of EPC histologically confirmed?10 (29·4)22 (64·7)2 (5·9)
Q6. Was the reference standard for diagnosis described in sufficient detail?24 (70·6)10 (29·4)0 (0·0)
Q7. Was the index test (K-ras mutational analysis) described in sufficient detail?34 (100)0 (0·0)0 (0·0)
Q8. Was the definitive diagnosis made without knowledge of K-ras mutational status?13 (38·2)2 (5·9)19 (55·9)
Q9. The K-ras results were interpreted without knowledge of the definitive diagnosis?9 (26·5)2 (5·9)23 (67·6)
Q10. Did K-ras form part of the reference standard?1 (2·9)33 (97·1)0 (0·0)
Q11. Did all patients receive the same reference standard regardless of their K-ras result?28 (82·4)0 (0·0)6 (17·6)
Q12. Is the time period between reference standard and K-ras analysis short enough to  be reasonably sure that the target condition had not changed?34 (100)0 (0·0)0 (0·0)
Q13. Was K-ras mutational status reported for all patients described as having  entered in the study?19 (55·9)15 (44·1)0 (0·0)
Q14. Were the motives for which some patients did not contribute to K-ras analysis explained?10 (66·7)5 (33·3)0 (0·0)

Diagnostic accuracy

The sensitivity that studies observed for detecting the mutational status of K-ras in the diagnosis of EPC ranged from 0 through 100%, while the specificity ranged from 58·7% through 100%. The heterogeneity present within the study estimates can be observed in the ROC diagram (Fig. 3). We attempted to explain this heterogeneity by exploring the study characteristics. Notably, the different types of samples used appeared to influence the diagnostic sensitivity a great deal (Table 3). The sensitivity of K-ras tests in serum or plasma samples, in stool samples and in pancreatic juice was less than adequate: 37·6% (17·5–63·1), 65·7% (38·2–85·5) and 60·0 (43·6–74·5), respectively. Thus, we focused our analysis on the 17 studies that were carried out using tissue samples. The mean sensitivity and specificity in this group was 76·5% (66·7–84·2) and 91·8% (87·6–94·7), respectively. We sought to evaluate the possible causes of heterogeneity using bivariate meta-regression.

image

Figure 3.  Meta-analysis of 34 studies evaluating the diagnostic utility of detecting K-ras mutations in the diagnosis of exocrine pancreatic cancer in at least 50 patients.

Download figure to PowerPoint

Table 3.   Sensitivity and specificity of detecting K-ras mutations for the diagnosis of exocrine pancreatic cancer. Summary of results of the 34 studies, according to type of sample and authors’ conclusions
 Number of studies (n = 34) N (%)Sensitivity mean (95% CI)Specificity mean (95% CI)
Type of sample
 Tissue17 (50·0)76·5 (66·7–84·2)91·8 (87·6–94·7)
 Pancreatic juice, bile or other digestive secretions9 (26·5)60·0 (43·6–74·5)83·6 (76·2–89·0)
 Serum or plasma4 (11·8)37·6 (17·5–63·1)94·0 (86·3–97·5)
 Stool3 (8·8)65·7 (38·2–85·5)75·8 (59·1–87·2)
 Various1 (2·9)76·1 (29·4–96·1)99·1 (84·1–100)
Patient selection design
 Case–control19 (55·8)72·5 (60·9–81·7)88·1 (83·1–91·8)
 Suspicion of exocrine pancreatic cancer4 (11·8)68·3 (42·3–86·4)94·1 (83·8–98·0)
 Suspicion of biliary or pancreatic disease3 (8·8)68·3 (38·4–88·1)91·1 (78·3–96·6)
 Patients undergoing a certain procedure5 (14·7)56·5 (33·2–77·2)93·2 (84·0–97·3)
 Unclear3 (8·8)52·6 (22·7–80·8)77·4 (60·8–88·3)
Authors’ conclusion regarding clinical application
 Favourable as an independent diagnostic procedure13 (38·2)69·9 (55·8–81·0)90·3 (84·4–94·1)
 Favourable as an adjunct to another diagnostic procedure13 (38·2)73·9 (61·3–83·5)87·2 (79·6–92·3)
 Favourable only in a limited context3 (8·8)61·4 (33·0–83·7)90·7 (74·6–97·0)
 Not favourable in any context5 (14·7)45·8 (24·1–69·3)88·4 (76·6–94·7)

Table 4 shows that the sensitivity appeared to be slightly higher when K-ras was determined in histological samples than in cytological samples, but this difference may have occurred by chance (P = 0·166). Furthermore, the sensitivity was highest in studies that used a case–control design, reporting 82·7% (73·3–88·7). This difference was statistically significant when compared to the other more clinically relevant patient groups (82·7% vs. 68·4%, P = 0·039). Whether studies reported consecutive recruitment or not did not appear to influence diagnostic accuracy. Studies reporting that the K-ras mutational status had been interpreted without knowledge of the definitive diagnosis had slightly lower sensitivity than those that were not interpreted blindly (although the difference was not statistically significant, 67·5% vs. 77·1%, P = 0·286), and studies that determined K-ras in all patients described as having entered the study reported somewhat higher sensitivity than studies that were unable to report K-ras results for some patients (83·8% vs. 66·3%, P = 0·004).

Table 4.   Sensitivity and specificity of detecting K-ras mutations in tissue samples for the diagnosis of exocrine pancreatic cancer. Summary of results of the 17 studies, according to study characteristics and methodological quality
 Number of studies (n = 17) N (%)Sensitivity mean (95% CI)Specificity mean (95% CI)
Type of tissue sample
 Cytology6 (35·3)72·1 (59·6–81·9)91·9 (84·6–95·9)
 Histology8 (47·1)83·0 (70·7–90·8)92·0 (83·6–96·3)
 Unclear2 (11·8)73·8 (39·5–92·4)96·9 (57·3–99·9)
 Mixed1 (5·9)70·8 (46·1–87·3)95·0 (73·2–99·2)
Patient selection design
 Case–control8 (47·1)82·7 (73·3–88·7)92·4 (85·7–96·0)
 Suspicion of exocrine pancreatic cancer4 (23·5)68·5 (53·3–80·5)94·2 (84·1–98·0)
 Suspicion of biliary or pancreatic disease2 (11·8)81·0 (62·6–91·5)93·2 (72·8–98·6)
 Patients undergoing a certain procedure2 (11·8)54·8 (34·3–73·9)95·6 (80·4–99·2)
 Unclear1 (5·9)79·6 (52·0–93·3)79·6 (49·8–93·9)
Patient selection design
 Case–control8 (47·1)82·7 (73·0–89·4)92·2 (85·8–95·8)
 More clinically relevant populations8 (47·1)68·4 (57·0–78·0)94·3 (88·2–97·3)
 Unclear1 (5·9)79·6 (49·1–94·0)79·6 (52·2–93·3)
Consecutive patient selection
 Yes6 (35·3)74·6 (61·3–84·4)93·7 (85·3–97·4)
 No11 (64·7)76·3 (66·5–84·0)91·7 (85·9–95·2)
Q1. Was the spectrum of patients representative of the patients who will receive the test in practice?
 Yes5 (29·4)76·2 (62·2–86·2)93·5 (83·9–97·6)
 No/unclear12 (70·6)75·3 (66·0–82·7)91·9 (86·5–95·3)
Q5. Were all of cases of EPC histologically confirmed?
 Yes5 (29·4)76·9 (62·2–87·5)95·3 (89·4–98·0)
 No/unclear12 (70·6)75·1 (66·1–82·3)89·9 (84·1–93·8)
Q8. The definitive diagnosis was made without knowledge of K-ras mutational status?
 Yes9 (52·9)79·3 (61·4–87·5)95·3 (89·4–98·0)
 No/unclear8 (47·1)72·3 (66·1–82·3)89·9 (84·1–93·8)
Q9. The K-ras results were interpreted without knowledge of the definitive diagnosis?
 Yes3 (17·6)67·5 (48·3–82·2)91·5 (85·2–95·2)
 No/unclear14 (82·4)77·1 (69·4–83·4)93·5 (86·1–97·1)
Q11. Did all patients receive the same reference standard regardless of their K-ras result?
 Yes14 (82·4)78·3 (71·8–83·7)91·4 (86·4–94·7)
 No/unclear3 (17·6)60·0 (43·8–74·3)96·0 (84·7–99·0)
Q13. Was K-ras mutational status reported for all patients described as having entered in the study?
 Yes8 (47·1)83·8 (76·2–89·3)89·0 (82·2–93·4)
 No/unclear9 (52·9)66·3 (56·0–75·2)95·2 (90·0–97·8)
Q14. Were the motives for which some patients did not contribute to K-ras analysis explained?
 Yes5 (62·5)65·0 (51·8–76·3)96·3 (89·3–98·8)
 No/unclear3 (37·5)69·1 (50·6–83·0)94·0 (83·6–98·0)

Only five studies reported K-ras mutational status by tumour stage. In all of them, the sensitivity of the test in stage IV tumours was greater than in stages I–III (Table S7). The pooled estimate was 30·5% (19·5–44·0) for tumours in stages I–III and 71·6% (61·6–79·8) for stage IV tumours.

Authors’ conclusions

Five studies (14·7%) were considered to be unfavourable to the diagnostic application of K-ras in any context (Table 3). Thirteen studies (38·2%) favoured the clinical utility of detecting K-ras as an independent diagnostic procedure, 13 studies (38·2%) were favourable to testing K-ras only in combination with another diagnostic procedure, and three additional studies (8·8%) were favourable to K-ras tests only in certain diagnostic contexts. For all studies included in this review, an extracted text regarding the conclusions of the authors of the studies on the clinical application can be found in the Table S3, along with how we graded the authors’ conclusions.

With regard to the utility of detecting K-ras in combination with another diagnostic procedure, it was possible to calculate the diagnostic accuracy of K-ras in combination with cytopathology in eight studies. In several of them, the sensitivity was close to 90% (Table 5). In one study [47], the authors reported a lower diagnostic accuracy in patients without a pancreatic mass. We also extracted information regarding the sensitivity and specificity of K-ras in combination with tumour marker CA 19-9, with other genetic markers and other combinations of tests; detailed information can be found in the Tables S8–S10.

Table 5.   Sensitivity and specificity of detecting K-ras mutations in combination with cytopathology in 8 studies with available data
StudySensitivity (95% CI)*Specificity (95% CI)*
  1. NR, Not reported.

  2. *When confidence intervals were not reported, we used the numbers available in the studies. In some instances, this was not possible because raw patient numbers were not provided.

  3. Group 1 – suspected EPC with pancreatic masses.

  4. Group 2 – suspected EPC without pancreatic masses.

Maluf-Filho, 2007 [60]89·5 (81·5–97·4)94·1 (73·9–100)
Pelliséet al [48]97·0 (82·5–99·8)91·7 (71·5–98·5)
Zheng, 2003 [61]83·3 (70·2–91·6)100 (69·9–100)
Urgell et al [47]88·3 (76·8–94·8)100 (82·8–100)
Urgell et al [47]53·3 (27·4–77·7)61·5 (32·3–84·9)
Villanueva et al [49]77·6 (66·4–86·1)100 (77·1–100)
Mora, 1998 [62]91·3 (78·3–97·2)NR
Van Laethem, 1998 [63]8675

As mentioned, three studies were favourable to K-ras tests only in certain contexts: two studies suggested that K-ras should be performed only when the cytological analysis is not conclusive [48,49] and the other study suggested that K-ras may prove to be useful in patients who have normal levels of CA 19-9, but in whom EPC remains suspected [50].

If we compare the diagnostic accuracy achieved in the studies with the authors’ conclusion, we show that studies that were unfavourable towards the clinical application of K-ras reported lower diagnostic accuracy, as would be expected (Table 3). There did not appear to be substantial differences in the diagnostic accuracy reported in the other three categories of authors’ conclusions.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure of interests
  9. Address
  10. References
  11. Supporting Information

We found and assessed 34 studies evaluating the diagnostic utility of analysing the K-ras mutational status in EPC in at least 50 patients. The studies showed numerous methodological limitations, a broad range of diagnostic accuracy values and heterogeneous conclusions in relation to the clinical application of this potential diagnostic test. The methodological quality of the 34 studies was rather poor, and many of the studies lacked vital information required to determine the external validity. Alarmingly, almost one in four studies did not even report the age and sex of the patients included. Nevertheless, over half of the studies stated to be favourable to the independent application of K-ras, while a number of studies concluded that K-ras mutations could be useful in the diagnosis of EPC when combined with cytohistopathology. Only a limited number of studies allowed us to evaluate the combined sensitivity and specificity of K-ras and other techniques; in them, we did observe higher values of diagnostic sensitivity. It is possible that these studies reported the combined sensitivity and specificity because it was higher; thus, this finding could represent selective reporting bias. It is difficult to further assess this issue (e.g., without access to the original study protocol).

If we consider K-ras analysis as an independent diagnostic procedure, we can conclude that, as expected, accuracy is highest when K-ras is detected in cytohistological material: the mean sensitivity and specificity were 76·5% and 91·8%, respectively. The sensitivity of K-ras detection in other biological samples, such as stool, blood or pancreatic juice, did not appear to be appropriate for diagnostic use. Even in tissue samples, it is possible that the mean values of sensitivity and specificity that we summarize are optimistic because of methodological limitations (such as reviewer bias or spectrum bias) of the reviewed studies. Notably, the mean sensitivity dropped below 70% when K-ras was interpreted without knowledge of the reference standard or when studies were carried out in clinically relevant patient populations rather than within case–control designs. Classically, case–control designs in diagnostic research have been shown to have a tendency to inflate diagnostic accuracy [18,20,24,30], as we see here. Furthermore, in some studies with a high degree of validity, the prevalence at diagnosis of K-ras mutations has been similar in cytological and histological samples [51,52].

The sensitivity was lower in studies that were unable to determine K-ras in all initially selected patients. This might reflect some selection bias [29,53]: the eligible patients in whom K-ras was not determined could be patients with more advanced pancreatic cancer and, perhaps, with more frequently mutated K-ras (although the prevalence of mutations has been found to be unrelated to stage in other large studies) [54]. It is also possible that several studies that apparently reported K-ras results for all patients included actually only reported the inclusion of subjects with K-ras results. Notably, many of the studies were performed before publication of the STARD statement [55], which encourages a more transparent reporting of patient eligibility and inclusion.

This brings us on to one limitation of the present kind of research: evaluating associations between study characteristics and outcomes relies on transparent and good-quality reporting of research methodology. Unfortunately, despite increased sensitization to issues related to the quality of reporting, diagnostic research remains poorly reported [28,33,55]. For example, patient recruitment was considered unclear in three studies, and 22 (64·7%) did not fully explain the reasons for inclusion as required by STARD [55]. Furthermore, it is unlikely that we could include all studies that addressed the utility of K-ras for EPC diagnosis; for example, research articles that study K-ras as part of a larger panel of molecular markers may not have been detected by our search strategy. While we only included published studies, detection of all diagnostic studies in electronic databases remains challenging. However, the studies described here represent a substantial and systematically obtained sample that can adequately address the objectives of the present study: to assess the clinical utility of K-ras and to understand how certain procedural or methodological characteristics influence the diagnostic accuracy.

Although detection of K-ras mutations proved fairly specific in clinically relevant populations, the difficulties in procuring appropriate biological samples may limit its clinical utility. The mean sensitivity of detecting K-ras mutations in patients with suspected EPC – the most clinically relevant population – was 68·3%. This would lead to one in three cases of EPC being missed if decisions were based solely on K-ras mutational status. Given that the highest diagnostic accuracy was observed when K-ras was detected in cytohistological samples, it is possible that the two procedures should be part of future diagnostic strategies.

More than 35 years after the discovery of the K-ras oncogene, the relevance of K-ras mutations in carcinogenesis is well established [56–59]; it is hence remarkably unfortunate that their role in the diagnosis of patients suffering from pancreatic and other cancers remains unclear. An adequate validation following the rules of evidence-based diagnosis is crucial to discover clinically meaningful tumour markers. The methodological flaws seen in diagnostic validation of biomarkers with well-established biological properties and roles, as K-ras, become even more relevant when the promises of ‘personalized medicine’ are pondered [27,40].

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure of interests
  9. Address
  10. References
  11. Supporting Information

Supported in part by research grants from CIBER de Epidemiología y Salud Pública (CIBERESP), Instituto de Salud Carlos III, Madrid, Government of Spain; and the Government of Catalonia (2009 SGR 1350).

Disclosure of interests

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure of interests
  9. Address
  10. References
  11. Supporting Information

The authors declare they have no competing financial interests. The study sponsors had no role and no involvement in the study design nor in the collection, analysis and interpretation of data; they also had no role in the writing of the report nor in the decision to submit the paper for publication.

Address

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure of interests
  9. Address
  10. References
  11. Supporting Information

Department of Public Health, Miguel Hernández University, Alicante, Spain (L. A. Parker, B. Lumbreras, I. Hernández-Aguado); CIBER en Epidemiología y Salud Pública (CIBERESP) (L. A. Parker, B. Lumbreras, I. Hernández-Aguado, M. Porta); Institut Municipal d’Investigació Mèdica, Barcelona, Catalonia, Spain (T. Lopez, M. Porta); School of Medicine, Universitat Autònoma de Barcelona, Spain (M. Porta).

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure of interests
  9. Address
  10. References
  11. Supporting Information
  • 1
    Ferlay J, Autier P, Boniol M, Heanue M, Colombet M, Boyle P. Estimates of the cancer incidence and mortality in Europe in 2006. Ann Oncol 2007;18:58192.
  • 2
    Fernandez E, La Vecchia C, Porta M, Negri E, Lucchini F, Levi F. Trends in pancreatic cancer mortality in Europe, 1955–1989. Int J Cancer 1994;57:78692.
  • 3
    Stathis A, Moore MJ. Advanced pancreatic carcinoma: current treatment and future challenges. Nat Rev Clin Oncol 2010;7:16372.
  • 4
    Porta M, Fabregat X, Malats N, Guarner L, Carrato A, de Miguel A et al. Exocrine pancreatic cancer: symptoms at presentation and their relation to tumour site and stage. Clin Transl Oncol 2005;7:18997.
  • 5
    Michl P, Pauls S, Gress TM. Evidence based diagnosis and staging of pancreatic cancer. Clin Gastroenterol 2006;20:22751.
  • 6
    Fry LC, Mönkemüller K, Malferthereiner P. Molecular markers of pancreatic cancer: development and clinical relevance. Langenbecks Arch Surg 2008;393:88390.
  • 7
    Porta M, Malats N, Piñol JL, Rifà J, Andreu M, Real FX. Diagnostic certainty and potential for misclassification in exocrine pancreatic cancer. J Clin Epidemiol 1994;47:106979.
  • 8
    Duffy MJ, Sturgeon C, Lamerz R, Haglund C, Holubec VL, Klapdor R et al. Tumor markers in pancreatic cancer: a European Group on Tumor Markers (EGTM) status report. Ann Oncol 2010;21:4417.
  • 9
    Almoguera C, Shibata D, Forrester K, Martin J, Arnheim N, Perucho M. Most human carcinomas of the exocrine pancreas contain mutant c-K-ras genes. Cell 1998;53:54954.
  • 10
    Tada M, Omata M, Ohto M. Clinical application of ras gene mutation for diagnosis of pancreatic adenocarcinoma. Gastroenterology 1991;100:2338.
  • 11
    Malats N, Porta M, Corominas JM, Piñol JL, Rifà J, Real FX. Ki-ras mutations in exocrine pancreatic cancer: associations with clinico-pathological characteristics and with tobacco and alcohol consumption. Int J Cancer 1997;70:6617.
  • 12
    Okai T, Watanabe H, Yamaguchi Y, Mouri I, Motoo Y, Sawabu N. EUS and K-ras analysis of pure pancreatic juice collected via a duodenoscope after secretin stimulation for diagnosis of pancreatic mass lesion: a prospective study. Gastrointest Endosc 1999;50:797803.
  • 13
    Trümper LH, Bürger B, von Bonin F, Hintze A, von Blohn G, Pfreundschuh M et al. Diagnosis of pancreatic adenocarcinoma by polymerase chain reaction from pancreatic secretions. Br J Cancer 1994;70:27884.
  • 14
    Castells A, Puig P, Móra J, Boadas J, Boix L, Urgell E. K-ras mutations in DNA extracted from the plasma of patients with pancreatic carcinoma: diagnostic utility and prognostic significance. J Clin Oncol 1999;17:57884.
  • 15
    Wenger FA, Zieren J, Peter FJ, Jacobi CA, Müller JM. K-ras mutations in tissue and stool samples from patients with pancreatic cancer and chronic pancreatitis. Langenbecks Arch Surg 1999;384:1816.
  • 16
    Berger DH, Chang H, Wood M, Huang L, Heath CW, Lehman T et al. Mutational activation of K-ras in nonneoplastic exocrine pancreatic lesions in relation to cigarette smoking status. Cancer 1999;85:32632.
  • 17
    GressTM (ed). Molecular Pathogenesis of Pancreatic Cancer. Amsterdam: IOS Press, 2000.
  • 18
    Lijmer JG, Mol BW, Heisterkamp S, Bonsel GJ, Prins MH, van der Meulen JH et al. Empirical evidence of design-related bias in studies of diagnostic tests. JAMA 1999;282:10616.
  • 19
    Ransohoff DF. Bias as a threat to the validity of cancer molecular-marker research. Nat Rev Cancer 2005;5:1429.
  • 20
    Porta M (ed). A Dictionary of Epidemiology, 5th edn. New York: Oxford University Press, 2008. p. 66, 69, 191, 201, 227, 233–6, 255, 258.
  • 21
    Feinstein AR. Clinical Epidemiology. The Architecture of Clinical Research. Philadelphia: WB Saunders; 1985.
  • 22
    Fletcher RH, Fletcher SW. Clinical Epidemiology. The Essentials, 4th edn. Philadelphia: Lippincott, Williams & Wilkins; 2005.
  • 23
    Haynes RB, Sackett DL, Guyatt GH, Tugwell P. Clinical Epidemiology. How to Do Clinical Practice Research, 3rd edn. Philadelphia: Lippincott, Williams & Wilkins; 2005.
  • 24
    Knotternuss JA, van Weel C, Muris JWM. Evidence base of clinical diagnosis: evaluation of diagnostic procedures. BMJ 2002;324:47780.
  • 25
    Hernández-Aguado I, Porta M, Miralles M, García Benavides F, Bolúmar F. La cuantificación de la variabilidad en las observaciones clínicas. Med Clin (Barc) 1990;95:4249.
  • 26
    Reid MC, Lachs MS, Feinstein AR. Use of methodological standards in diagnostic test research. Getting better but still not good enough. J Am Med Assoc 1995;274:64551.
  • 27
    Porta M, Hernández-Aguado I, Lumbreras B, Crous-Bou M. ‘Omics’ research, monetization of intellectual property and fragmentation of knowledge: can clinical epidemiology strengthen integrative research? J Clin Epidemiol 2007;60:12205.
  • 28
    Lumbreras B, Parker LA, Porta M, Pollán M, Ioannidis JPA, Hernández-Aguado I. Overinterpretation of clinical applicability in molecular diagnostic research. Clin Chem 2009;55:78694.
  • 29
    Hoppin JA, Tolbert PE, Taylor JA, Schroeder JC, Holly EA. Potential for selection bias with tumor tissue retrieval in molecular epidemiology studies. Ann Epidemiol 2002;12:16.
  • 30
    Whiting P, Rutjes AW, Reitsma JB, Glas AS, Bossuyt PM, Kleijnen J. Sources of variation and bias in studies of diagnostic accuracy: a systematic review. Ann Intern Med 2004;140:189202.
  • 31
    Sackett DL, Haynes RB. Evidence base of clinical diagnosis: the architecture of diagnostic research. BMJ 2002;324:53941.
  • 32
    Pepe MS, Etzioni R, Feng Z, Potter JD, Thompson ML, Thornquist M et al. Phases of biomarker development for early detection of cancer. J Natl Cancer Inst 2001;93:105461.
  • 33
    Lumbreras B, Porta M, Márquez S, Pollán M, Parker LA, Hernández-Aguado I. Sources of error and its control in studies on the diagnostic accuracy of ‘-omics’ technologies. Proteomics Clin Applic 2009;3:17384.
  • 34
    Rutjes AW, Reitsma JB, Vandenbroucke JP, Glas AS, Bossuyt PM. Case-control and two-gate designs in diagnostic accuracy studies. Clin Chem 2005;51:133541.
  • 35
    Porta M, Ferrer-Armengou O, Pumarega J, López T, Crous-Bou M, Alguacil J et al. Exocrine pancreatic cancer clinical factors were related to timing of blood extraction and influenced serum concentrations of lipids. J Clin Epidemiol 2008;61:695704.
  • 36
    Porta M, Malats N, Piñol JL, Real FX. Re.: diagnostic certainty in pancreatic cancer. Relevance of misclassification of disease status in epidemiologic studies of exocrine pancreatic cancer. Response to Silverman et al. J Clin Epidemiol 1996;49:6023.
  • 37
    Porta M, Pumarega J, Ferrer-Armengou O, López T, Alguacil J, Malats N et al. Timing of blood extraction in epidemiologic and proteomic studies: results and proposals from the PANKRAS II Study. Eur J Epidemiol 2007;22:57788.
  • 38
    Soler M, Porta M, Malats N, Guarner L, Costafreda S, Gubern JM et al. Learning from case reports: diagnostic issues in an epidemiologic study of pancreatic cancer. J Clin Epidemiol 1998;51:121521.
  • 39
    Porta M, Costafreda S, Malats N, Guarner L, Soler M, Gubern JM et al. Validity of the hospital discharge diagnosis in epidemiologic studies of biliopancreatic pathology. Eur J Epidemiol 2000;16:53341.
  • 40
    Ioannidis JP. Expectations, validity, and reality in omics. J Clin Epidemiol 2010;63:9459.
  • 41
    Whiting P, Rutjes AW, Reitsma JB, Bossuyt PM, Kleijnen J. The development of QUADAS: a tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews. BMC Med Res Methodol 2003;3:25.
  • 42
    Harbord RM (2008) Metandi: Stata Module for Meta-Analysis of Diagnostic Accuracy. Chestnut Hill, MA: Statistical Software Components, Boston College Department of Economics. Revised 15 Apr 2008.
  • 43
    Chu H, Cole SR. Bivariate meta-analysis of sensitivity and specificity with sparse data: a generalized linear mixed model approach. J Clin Epidemiol 2006;59:13312.
  • 44
    Reitsma JB, Glas AS, Rutjes AW, Scholten RJ, Bossuyt PM, Zwinderman AH. Bivariate analysis of sensitivity and specificity produces informative summary measures in diagnostic reviews. J Clin Epidemiol 2005;58:98290.
  • 45
    Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA Statement. PLoS Med 2009;6:e1000097.
  • 46
    Simera I, Moher D, Hoey J, Schulz KF, Altman DG. A catalogue of reporting guidelines for health research. Eur J Clin Inves 2010;40:3553.
  • 47
    Urgell E, Puig P, Boadas J, Capellà G, Queraltó JM, Boluda R et al. Prospective evaluation of the contribution of K-ras mutational analysis and CA 19.9 measurement to cytological diagnosis in patients with clinical suspicion of pancreatic cancer. Eur J Cancer 2000;36:206975.
  • 48
    Pellisé M, Castells A, Ginès A, Solé M, Mora J, Castellví-Bel S et al. Clinical usefulness of KRAS mutational analysis in the diagnosis of pancreatic adenocarcinoma by means of endosonography-guided fine-needle aspiration biopsy. Aliment Pharmacol Ther 2003;17:1299307.
  • 49
    Villanueva A, Reyes G, Cuatrecasas M, Martínez A, Erill N, Lerma E et al. Diagnostic utility of K-ras mutations in fine-needle aspirates of pancreatic masses. Gastroenterology 1996;110:158794.
  • 50
    Maire F, Micard S, Hammel P, Voitot H, Lévy P, Cugnenc PH et al. Differential diagnosis between chronic pancreatitis and pancreatic cancer: value of the detection of KRAS2 mutations in circulating DNA. Br J Cancer 2002;87:5514.
  • 51
    Porta M, Malats N, Guarner L, Carrato A, Rifà J, Salas A et al. Association between coffee drinking and K-ras mutations in exocrine pancreatic cancer. J Epidemiol Community Health 1999;53:7029.
  • 52
    Parker LA, Porta M, Lumbreras B, López T, Guarner L, Hernández-Aguado I et al. Clinical validity of detecting K-ras mutations for the diagnosis of exocrine pancreatic cancer: a prospective study in a clinically-relevant spectrum of patients. Eur J Epidemiol 2011; doi: 10.1007/s10654-011-9547-8 (in press).
  • 53
    Porta M, Malats N, Corominas JM, Rifâ J, Piñol JL, Real FX et al. Generalising molecular results arising from incomplete biological samples: expected bias and unexpected findings. Ann Epidemiol 2002;12:714.
  • 54
    Porta M, Malats N, Vioque J, Carrato C, Soler M, Ruiz L et al. Incomplete overlapping of biological, clinical and environmental information in molecular epidemiologic studies: a variety of causes and a cascade of consequences. J Epidemiol Community Health 2002;56:7348.
  • 55
    Smidt N, Rutjes AW, van der Windt DA, Ostelo RW, Bossuyt PM, Reitsma JB et al. The quality of diagnostic accuracy studies since the STARD statement: has it improved? Neurology 2006;67:7401.
  • 56
    Malumbres M, Barbacid M. RAS oncogenes: the first 30 years. Nat Rev Cancer 2003;3:45965.
  • 57
    Croce CM. Oncogenes and cancer. N Engl J Med 2008;358:50211.
  • 58
    Porta M, Ayude D, Alguacil J, Jariod M. Exploring environmental causes of altered ras effects: fragmentation + integration? Mol Carcinog 2003;36:4552.
  • 59
    Porta M, Crous-Bou M, Wark PA, Vineis P, Real FX, Malats N et al. Cigarette smoking and K-ras mutations in pancreas, lung and colorectal adenocarcinomas: etiopathogenic similarities, differences and paradoxes. Mutat Res Rev 2009;682:8393.
  • 60
    Maluf-Filho F, Kumar A, Gerhardt R, Kubrusly M, Sakai P, Hondo F et al. Kras mutation analysis of fine needle aspirate under EUS guidance facilitates risk stratification of patients with pancreatic mass. J Clin Gastroenterol 2007;41:90610.
  • 61
    Zheng M, Liu LX, Zhu AL, Qi SY, Jiang HC, Xiao ZY. K-ras gene mutation in the diagnosis of ultrasound guided fine-needle biopsy of pancreatic masses. World J Gastroenterol 2003;9:18891.
  • 62
    Mora J, Puig P, Boadas J, Urgell E, Montserrat E, Lerma E et al. K-ras gene mutations in the diagnosis of fine-needle aspirates of pancreatic masses: prospective study using two techniques with different detection limits. Clin Chem 1998;44:22438.
  • 63
    Van Laethem JL, Bourgeois V, Parma J, Delhaye M, Cochaux P, Velu T et al. Relative contribution of Ki-ras gene analysis and brush cytology during ERCP for the diagnosis of biliary and pancreatic diseases. Gastrointest Endosc 1998;47:47985.

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Disclosure of interests
  9. Address
  10. References
  11. Supporting Information

Table S1. Search terms used for detecting original research articles that evaluated the diagnostic accuracy of detecting K-ras mutation for diagnosis of EPC.

Table S2. Stated objectives of 30 articles evaluating the utility of assessing the K-ras mutational status for the diagnosis of exocrine pancreatic cancer in at least 50 patients.

Table S3. Conclusions of 30 articles regarding the clinical application of detecting K-ras mutations in the diagnosis of exocrine pancreatic cancer in at least 50 patients.

Table S4. Data extracted from 34 patient series evaluating the utility of detecting K-ras mutations in the diagnosis of exocrine pancreatic cancer in at least 50 patients.

Table S5. Summary of the analytical test characteristics in 30 articles evaluating K-ras mutations for the diagnosis of exocrine pancreatic cancer in at least 50 patients

Table S6. Quality analysis of 34 patient series evaluating the utility of detecting K-ras mutations in the diagnosis of exocrine pancreatic cancer in at least 50 patients

Table S7. Diagnostic sensitivity of K-ras mutations in exocrine pancreatic cancer according to cancer stage

Table S8. Sensitivity and specificity of detecting K-ras mutations in combination with tumour marker CA19·9 in 3 studies with available data.

Table S9. Sensitivity and specificity of detecting K-ras mutations in combination with other genetic markers.

Table S10. Sensitivity and specificity of detecting K-ras mutations in combination with other diagnostics procedures.

Annex. References of 30 articles (34 studies) evaluating the utility of assessing the K-ras mutational status for the diagnosis of exocrine pancreatic cancer in at least 50 patients.

FilenameFormatSizeDescription
ECI_2495_sm_TableS1-S10.doc316KSupporting info item
ECI_2495_sm_Annex.doc64KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.