Technical and clinical performance of a new assay to

Download PDF
0 downloads 0 Views 833KB Size Report
Comment
4Xuanwu Hospital, Capital Medical University, Beijing Shi, China. 5Roche ...... Ernst JD, et al. Murray &. Nadel's textbook of respiratory medicine. 6th ed. Philadel- .... classification of malignant tumours. 7th ed. Hoboken: Wiley-Blackwell, 2009. 42. ... fessionals/physician_gls/PDF/cervical.pdf (2017, accessed. 31 March 2017) ...
Original Article

Technical and clinical performance of a new assay to detect squamous cell carcinoma antigen levels for the differential diagnosis of cervical, lung, and head and neck cancer

Tumor Biology April 2018: 1–13 Ó The Author(s) 2018 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/1010428318772202 journals.sagepub.com/home/tub

Stefan Holdenrieder1, Rafael Molina2, Ling Qiu3, Xiuyi Zhi4, Sandra Rutz5, Christine Engel5, Pia Kasper-Sauer6, Farshid Dayyani7,8 and Catharina M Korse9

Abstract In squamous cell carcinoma, squamous cell carcinoma antigen levels are often elevated. This multi-center study evaluated the technical performance of a new ElecsysÒ squamous cell carcinoma assay, which measures serum squamous cell carcinoma antigen 1 and 2 levels in an equimolar manner, and investigated the potential of squamous cell carcinoma antigen for differential diagnosis of cervical, lung, and head and neck squamous cell carcinoma.Assay precision and method comparison experiments were performed across three European sites. Reference ranges for reportedly healthy individuals were determined using samples from banked European and Chinese populations. Differential diagnosis experiments determined whether cervical, lung, or head and neck cancer could be differentiated from apparently healthy, benign, or other malignant cohorts using squamous cell carcinoma antigen levels alone. Squamous cell carcinoma antigen cut-off levels were calculated based on squamous cell carcinoma antigen levels at 95% specificity. Repeatability coefficients of variation across nine analyte concentrations were  5.3%, and intermediate precision coefficients of variation were  10.3%. Method comparisons showed good correlations with Architect and Kryptor systems (slopes of 1.1 and 1.5, respectively). Reference ranges for 95th percentiles for apparently healthy individuals were 2.3 ng/mL (95% confidence interval: 1.9– 3.8; European cohort, n = 153) and 2.7 ng/mL (95% confidence interval: 2.2–3.3; Chinese cohort, n = 146). Strongest differential diagnosis results were observed for cervical squamous cell carcinoma: receiver operating characteristic analysis showed that squamous cell carcinoma antigen levels (2.9 ng/mL cut-off) differentiate cervical squamous cell carcinoma (n = 127) from apparently healthy females (n = 286; area under the curve: 86.2%; 95% confidence interval: 81.8–90.6; sensitivity: 61.4%; specificity: 95.6%), benign diseases (n = 187; area under the curve: 86.3%; 95% confidence interval: 81.2– 91.3; sensitivity: 61.4%; specificity: 95.0%), and other cervical cancers (n = 157; area under the curve: 78.9%; 95% confidence interval: 70.8–87.1; sensitivity: 61.4%; specificity: 86.7%). Squamous cell carcinoma may also aid in the differential diagnosis of lung cancer. The Elecsys squamous cell carcinoma assay exhibited good technical performance and is suitable for differential diagnosis of cervical squamous cell carcinoma in clinical practice. Keywords Squamous cell carcinoma, immunoassay, cervical, lung, head, neck

Date received: 6 September 2017; accepted: 20 February 2018

Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License (http://www.creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/ open-access-at-sage).

2

Introduction Squamous cell carcinoma (SCC) is an epithelial malignancy that can arise from different types of tissue, including lung, cervix, and head and neck.1 Due to the widespread location of epithelial lining in the respiratory and gynecological tracts, the prevalence of SCC is high in these regions. In non–small cell lung cancer (NSCLC), which accounts for approximately 85% of all patients with lung cancer, about 35% of cases are SCC.2,3 Furthermore, approximately 80% of cervical carcinomas and 90% of head and neck cancer cases are SCC.4,5 Squamous cell carcinoma antigen (SCCA) was first isolated from the uterus and is present in two isoforms: SCCA1 and SCCA2.6 Both isoforms are co-expressed in the squamous epithelium of the cervix, tongue, tonsils, esophagus, and vagina.7 Both isoforms of SCCA have been found to be elevated in cervical cancer.8,9 Levels are also elevated in the blood of patients with SCC of, for example, the cervix,9,10 lung,11 and head and neck.12,13 SCCA levels have also been shown to correlate with tumor stage in all of these SCC types.9,10,12–16 Also in SCC of the lung, head and neck, SCCA1 and SCCA2 are co-expressed in moderately and well-differentiated tumors.7,8 Therefore, for optimal clinical sensitivity, assays should be capable of detecting both SCCA1 and SCCA2 in order to determine the total SCCA present.13 Currently available assay systems are unable to differentiate the two isoforms in this manner, as they were not developed for this specific purpose. In cases of suspected disease, SCCA levels are measured in the patient’s serum using currently available immunoassays.17–20 As SCCA is present in saliva, hair, and skin particles that can be easily distributed in aerosols and dust, there is a high contamination risk that may lead to falsely elevated SCCA values in current assays.21 In addition, false positives are common in patients with benign renal, skin, lung, and liver diseases.12,22–24 SCCA levels are also sensitive to blood

1

University Hospital Bonn, Bonn, Germany Hospital Clinic de Barcelona and University of Barcelona, Barcelona, Spain 3 Peking Union Medical College Hospital, Beijing, China 4 Xuanwu Hospital, Capital Medical University, Beijing Shi, China 5 Roche Diagnostics GmbH, Penzberg, Germany 6 Roche Diagnostics GmbH, Mannheim, Germany 7 Roche Diagnostics International Ltd, Rotkreuz, Switzerland 8 University of California, Irvine, Irvine, CA, USA 9 The Netherlands Cancer Institute, Amsterdam, The Netherlands 2

Corresponding author: Catharina M Korse, The Netherlands Cancer Institute, PO Box 90203, Plesmanlaan 121, 1066 CX Amsterdam, The Netherlands. Email: [email protected]

Tumor Biology collection timing (before vs after anesthesia) and procedure (venous vs arterial vessel puncture).25 Previous studies have focused on the utility of SCCA as a marker for prognosis, monitoring, and recurrence in cervical,26–31 lung,32–34 and head and neck cancers.35,36 Some publications suggest that the utility of SCCA is for histologic subtyping because significantly higher concentrations are found in patients with lung SCC or cervical SCC than in patients with adenocarcinoma of these sites.37 However, published evidence on the practical use of SCCA to differentiate SCC from other types of malignancies or benign diseases is limited and its role is still unclear.13 Guidance on SCCA diagnostic cut-offs and their relevant sensitivity and specificity are needed in order to aid clinical decisions. This multi-center study was performed to first evaluate the technical performance of a new SCC assay, which was developed to measure serum levels of both SCCA1 and SCCA2. The second aim was to determine the distribution pattern of SCCA in cervical, lung, and head and neck SCC and investigate whether the new assay can be used for their differential diagnosis.

Methods SCC assay The ElecsysÒ SCC assay is an electrochemiluminescence immunoassay (ECLIA) that measures SCCA levels from serum samples and is used on the Elecsys and cobas e analyzers. The assay uses two SCC-specific monoclonal antibodies that recognize both SCCA1 and SCCA2 human isoforms in an equimolar manner. It utilizes a biotin–streptavidin sandwich principle and the read-out is via electrochemiluminescence. A 15-mL sample is required, and the total assay turnaround time is 18 min.

Sites and instruments The technical and clinical performance of the SCC assay was evaluated between June 2014 and June 2015 across three European sites in Amsterdam, Barcelona, and Bonn. The clinical evaluation (reference range/differential diagnosis) took place at four European sites and two Chinese sites and followed a mainly retrospective case–control study design. For the European reference ranges, samples were measured from a banked collection in Kiel. Sites in China (Peking Union Medical College Hospital (PUMCH) and XuanWu Hospitals in Beijing) participated in the clinical evaluation only, which took place from April to June 2015. All measurements (both technical and clinical) using the Elecsys SCC assay were performed on either a cobas e 411 or a cobas e 601 analyzer.

Holdenrieder et al.

Compliance with ethical standards All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all individual participants included in the study.

Sample sources and handling All sample materials were stored after measurement and kept at 220°C or 280°C until the end of the study. Sample shipments were maintained at 220°C or 280°C. All procedures were performed to minimize sample contamination from external sources by using fresh pipette tips and vessels.

Technical assessment Precision. Roche Diagnostics provided all sample material for the precision experiments, which were performed at the three European sites according to Clinical and Laboratory Standards Institute (CLSI) EP 05-A3 guidelines.38 The repeatability experiment was completed in Amsterdam over 21 days. Method comparisons. The SCC assay was compared with the commercially available Abbott Architect and Brahms Kryptor assays. Serum samples were collected from routine clinical practice and kept frozen until measurement. In addition, Roche Diagnostics provided human serum samples of high SCCA concentration and spiked samples not available from routine clinical practice. For each comparison, Passing–Bablok regression was used to estimate the agreement between the two systems.39 A minimum of 100 samples were required to be analyzed by each center, according to the CLSI EP09-A3 guidelines.40

Clinical assessment Reference range determination. Healthy samples from a single site (Kiel) were used to determine reference ranges for the European cohort; serum samples were collected from a sample bank from apparently healthy individuals at the Kiel site, and SCCA levels were measured in Bonn, Amsterdam, and Barcelona. For the Chinese population reference ranges, samples from apparently healthy individuals were collected from the PUMCH and XuanWu Hospitals in Beijing. All samples were measured at the PUMCH site. Recruitment was based on health checks and a questionnaire, and final inclusion for the reference range population was based on clinical chemistry

3 parameters, including glucose, cholinesterase, creatinine, C-reactive protein (CRP), and hemoglobin levels, which were measured in Kiel. Individuals were excluded if these clinical chemistry results were abnormal. A minimum of 120 samples was required to determine the reference ranges for the 90% confidence intervals (CI), and 146 samples were needed for the 95% CI analyses. Differential diagnosis. The main focus of the clinical part of the study was to determine whether cervical, lung, or head and neck cancer could be differentiated from apparently healthy, benign, or other malignant cohorts using SCCA levels alone. Pseudonymized samples from local serum banks were collected and used for the differential diagnosis experiments in Barcelona, Bonn, Amsterdam, and Beijing. European samples were measured at their respective European collecting sites, and all the Chinese samples were measured at PUMCH. The availability of informed consent or waiver was checked during source data verification. Patient samples were not taken before surgery or anesthesia, as is routine procedure for sample banks. Patient sample size determination for the differential diagnosis experiments was calculated as the minimum number of samples required to allow for basic descriptive analyses. The main inclusion criteria for the apparently healthy differential diagnosis population were clinically apparently healthy with normal routine clinical chemistry results (including bilirubin, CRP, alanine aminotransferase, aspartate aminotransferase, and creatinine to exclude inflammation, kidney, or liver disease); no known history of tumor disease or severe benign lung disorders; no acute lung disorders; not receiving medication related to the excluded diseases/disorders; and documented smoking habits. Samples were not derived from blood donors and had to be obtained from a variety of ethnic groups. For the malignant cohorts, patients were required to have primary tumors but to have not yet received therapy or surgery. Tumor Nodes Metastasis (TNM) and International Union Against Cancer (UICC) staging information was required.41 General exclusion criteria were as follows: aged \18 years, current pregnancy, and history of other malignant disease. Because SCCA levels tend to be higher in advanced chronic kidney disease (CKD), samples with a glomerular filtration rate (GFR) of \30 mL/min/ 1.73 m2 (CKD stages 4 and 5) were excluded from all analyses, except for the analysis of benign renal diseases.23,24,42,43 The GFR threshold of 30 mL/min/ 1.73 m2 was chosen because patients with a lower GFR were found to have CKD stages 4 and 5 (i.e. advanced) disease.44 The effect of SCCA levels on age, gender,

4

Tumor Biology

Table 1. Precision results of the SCC assay across all three sites.a Sample

SCC1 SCC2 HSP 01 HSP 02 HSP 03 HSP 04 HSP 05 HSP 06 HSP 07

Mean (ng/mL)

2.2 22.7 0.6 1.5 0.2 35.6 60.4 3.5 35.0

Repeatability

Intermediate precision

Reproducibility

SD (ng/mL)

CV (%)

SD (ng/mL)

CV (%)

SD (ng/mL)

CV (%)

0.1 0.6 0.0 0.1 0.1 1.2 1.8 0.1 0.7

3.8 2.6 5.3 4.4 4.1 3.4 3.0 3.7 2.0

0.2 1.3 0.1 0.1 0.2 2.0 4.0 0.2 1.5

7.0 5.6 10.3 7.7 6.3 5.5 6.6 5.7 4.3

0.2 1.4 0.1 0.1 0.2 2.3 4.3 0.2 1.5

8.0 6.1 14.0 8.7 7.0 6.4 7.1 6.8 4.3

SCC: squamous cell carcinoma; SD: standard deviation; CV: coefficient of variation; HSP: human serum pool. a Three outliers were excluded from the analysis consistent with CLSI EP05-A3.

site, ethnicity, smoking status, and UICC staging was evaluated in the apparently healthy, benign, and malignant cohorts.

Statistical analysis Measurements were captured using the Windows-based Computer Aided Evaluation (WinCAEv) software.45 Clinical data were collected using Medrio and MACRO software. All statistical analyses using clinical and technical information were performed using R version 3.0.1 and Statistical Analysis Software (SAS) version 9.3. Technical experiments were analyzed with all data and after outlier exclusion/replacement. A consistent outlier identification and handling method was applied at all sites. For the precision experiments, repeatability, intermediate precision, and reproducibility were calculated using SAS (version 9.3) and R (version 3.0.1, additive package variance component analysis (VCA) version 1.0.6), and outlier detection was performed, all as described in the CLSI EP05-A3 guideline.38 Passing– Bablok regression analysis and Pearson correlation coefficients were used to assess the correlation of the method comparison studies in compliance with CLSI EP09-A3.36 For the method comparisons, no outlier detection method was applied. Reference ranges were measured and statistically analyzed in accordance with the International Federation of Clinical Chemistry and Laboratory Medicine’s recommendations for reference range determination.46 Descriptive statistics for patient characteristics. To calculate reference ranges and determination of SCCA levels in the different cohorts, non-parametric quantile CIs based on the methods of Hahn and Meeker47 were used. Receiver operating characteristic (ROC) curves and areas under the curve (AUCs) were calculated for the differential diagnosis analyses. Based on these

results, the percentage sensitivity at 95% specificity was calculated; this 95% specificity was chosen based on the methods examined in a previous publication on cervical cancer.48 Cut-offs were based on the SCCA levels at 95% specificity for the comparisons. Patient data from the apparently healthy and benign head and neck cohorts that originated from the Barcelona site were excluded from the ROC analyses because these patients were unexpectedly positive for SCCA. No clear root cause could be found for these elevated levels of SCCA, but internal experiments performed at the site determined that these were likely due to known contamination risks such as saliva and dust, as reported in previous publications.21

Results Technical assessment Precision. The reproducibility, intermediate precision, and repeatability results across all sites are presented in Table 1. For samples with a mean .1.8 ng/mL, the coefficient of variation (CV) should be used to assess precision: a CV  5% for .1.8–5 ng/mL and a CV  6.5% for .5–70 mg/mL are specified as acceptable for the intermediate precision and a CV  10% for .1.8–5 ng/mL is specified as acceptable for the reproducibility. Reproducibility performed over 21 days at the Amsterdam site resulted in a CV in the range 4.3%– 8.0%. The intermediate precision ranged from 4.3% to 7.0%. The repeatability CVs were  3.8%. For samples with a mean  1.8 ng/mL, the standard deviation (SD) should be used to assess precision and an SD  0.2 is specified as acceptable. The range for reproducibility and intermediate precision was SD 0.1–0.2 and repeatability was SD 0.0–0.1. Method comparison. The slope between the cobas e 411 versus Architect was calculated as 1.13 (95% CI:

Holdenrieder et al.

5

Figure 1. Squamous cell carcinoma antigen (SCCA) assay method comparison results. (a) The cobas e 411 versus Architect system using serum samples performed at the Barcelona site (n = 193). (b) The cobas e 601 versus Kryptor system performed using serum samples at the Amsterdam site (n = 180).

1.04–1.20) and the intercept was 0.15 (95% CI: 0.11– 0.37) (Figure 1(a)). Pearson’s correlation was r = 0.937. For the comparison of the cobas e 601 and Kryptor systems, the slope was 1.50 (95% CI: 1.44–1.54), the intercept was 0.48 (95% CI: 0.40–0.56), and Pearson’s correlation was r = 0.988 (Figure 1(b)).

Clinical assessment Reference range determination. A total of 153 apparently healthy European individuals were included in the reference range analysis (Table 2). The median concentration of SCCA in apparently healthy individuals was 1.1 ng/mL. The 95% and 97.5% quantiles were 2.3 ng/ mL (95% CI: 1.9–3.8) and 2.7 ng/mL (95% CI: 2.2– 4.4), respectively. For the Chinese reference range cohort, samples from 146 individuals were analyzed. The median SCCA levels were 1.1 ng/mL, the 95th percentile was 2.7 ng/mL (95% CI: 2.2–3.3), and the 97.5th percentile was 3.0 ng/mL (95% CI: 2.5–4.1) (Table 3). Differential diagnosis: patient population. Patient demographics from the differential diagnostic population are shown in Table 2. The mean age was 56.7 years (SD: 15.4) and the male-to-female ratio was almost equal. At least half of the patients were either current (34.7%) or past (15.6%) smokers. The effect of site, age, gender, ethnicities, and smoking on SCCA levels in different cohorts was investigated, but no major impact was found (data not shown). Males tended to have slightly higher SCCA levels than females in all cohorts, but the difference was marginal (interquartile range (IQR)

overlapped: healthy males, 1.0–1.7 ng/mL and healthy females, 0.8–1.5 ng/mL) and most likely not clinically relevant, although could be a consequence of the composition of the cohorts. Differential diagnosis: cervical cancer. Significantly higher median levels of SCCA were measured in cervical SCC patients (7.4 (IQR: 1.6–19.2) ng/mL) compared with apparently healthy females (1.06 (IQR: 0.8–1.6) ng/ mL), patients with benign gynecological diseases (1.2 (IQR 0.9–1.4) ng/mL), and patients with other cervical cancers (1.7 (IQR 1.4–1.9) ng/mL) (Table 3). ROC analysis showed that cervical SCC could be differentiated from apparently healthy females with an AUC of 86.2% (95% CI: 81.8–90.6), benign gynecological diseases with an AUC of 86.3% (95% CI: 81.2– 91.3), and other cervical cancers with an AUC of 78.9% (95% CI: 70.8–87.1) (Figure 2(a)). The sensitivity at 95% specificity for the comparisons between cervical SCC and healthy women was 61.4%, benign patients was 61.4%, and other cervical cancers was 55.9% (Table 4). The SCCA cut-off based on 95% specificity for all cervical differential diagnosis comparisons was 2.9 ng/mL. With this cut-off level, the sensitivity for detecting cervical SCC versus apparently healthy females was 61.4% and the specificity was 95.6% (Table 4). Thus, 95.6% of apparently healthy females had SCCA levels  2.9 ng/mL, while 61.4% of patients with cervical SCC had SCCA levels .2.9 ng/mL. Due to very similar cut-off levels for each indication giving 95% specificity for the comparison with benign cervical cancer and other cervical cancers, the 2.9 ng/mL cut-off

14 (9.6) 6 (4.1) 126 (86.3) –

29 (19) 17 (11.1) 107 (69.9) – – – – – – –

146 (100) – – –

– – 153 (100) –

– – – – – –

38.5 (13.6) 55 (38)

49.5 (17.3) 75 (49)

– – – – – –

708 (34.7) 318 (15.6) 478 (23.4) 535 (26.2)

479 (23.5) 1 (0.1) 844 (41.4) 715 (35.1)

56.7 (15.4) 1034 (50.7)

– – – – – –

17 (6.9) 6 (2.4) 135 (54.9) 88 (35.8)

146 (59.4) 0 (0.0) 50 (20.3) 50 (20.3)

41.5 (14.5) 87 (35.4)

– – – – – –

60 (13.5) 38 (8.6) 100 (22.5) 246 (55.4)

91 (20.5) 0 (0.0) 353 (79.5) 0 (0.0)

53.2 (17.5) 208 (46.9)

Benign (n = 444)

449 (56.0) 211 (26.3) 107 (13.3) 35 (4.4) 215 (100) – 21 (9.8) 19 (8.8) 131 (60.9) 39 (18.1) 5 (23.2)

207 (25.8) 0 (0.0) 129 (16.1) 466 (58.1)

62.3 (11.4) 498 (62.1)

Lung cancer (n = 802)

SD: standard deviation; CKD: chronic kidney disease; SCC: squamous cell carcinoma. a Patients with CKD 4 and 5 (except for patients with benign renal disease) and the Barcelona apparently healthy cohort were excluded. b Staging information is for SCC patients only.

Age, mean year (SD) Male, n (%) Ethnicity, n (%) Asian-Chinese Black Caucasian/White No specification Smoking habits, n (%) Current Past Never No specification Staging, n (%)b 0 I II III IV N/A

Apparently healthy (n = 246)

All (N = 2039)

European (n = 153)

Chinese (n = 146)

Differential diagnosis population

Reference range population

Table 2. Patient demographics of the reference and differential diagnostic populations.a

86 (53.4) 19 (11.9) 14 (8.8) 41 (25.6) 154 (100) 10 (6.5) 14 (9.1) 13 (8.4) 24 (15.6) 92 (59.7) 1 (0.6)

0 (0.0) 1 (0.6) 135 (84.4) 24 (15.0)

60.8 (10.7) 121 (75.6)

Head and neck cancer (n = 160)

37 (23.6) 8 (5.1) 50 (31.9) 62 (39.5) 127 (100) – 31 (24.4) 45 (35.4) 34 (26.8) 16 (12.6) 1 (0.8)

4 (2.6) 0 (0.0) 110 (70.1) 43 (27.4)

51.6 (14.5) –

Cervical cancer (n = 157)

– – – – – –

59 (25.6) 36 (15.7) 72 (31.3) 63 (27.4)

31 (13.5) 0 (0.0) 67 (29.1) 132 (57.4)

61.1 (13.7) 120 (52.2)

Other malignancy (n = 230)

6 Tumor Biology

Holdenrieder et al.

7

Table 3. SCCA levels in the different cohorts. Reference range population

N

Median SCCA (ng/mL)

95% quantile (95% CI), 97.5% quantile (95% CI) (ng/mL)

European cohort Chinese cohort

153 146

1.1 1.1

2.3 (1.9–3.8), 2.7 (2.2–4.4) 2.7 (2.2–3.3), 3.0 (2.5–4.1)

Differential diagnosis population

N

Median SCCA (ng/mL)

IQR (ng/mL)

Apparently healthy Apparently healthy females Apparently healthy males Cervical SCC Adenocarcinoma Other Lunga NSCLC all NSCLC SCC NSCLC adenocarcinoma NSCLC excluding mixed/not otherwise specified SCLC all Other Head and neck SCC Adenocarcinoma Other Benign Cervical Lung Head and neck Other Renal Malignant other

246 159 87 157 127 18 12 802 608 215 261 18 189 5 160 154 2 4 444 60 123 69 192 44 290

1.2 1.1 1.3 3.5 7.4 1.2 1.7 1.2 1.4 1.9 1.0 1.4 1.0

0.8–1.6 0.8–1.6 1.0–1.8 1.4–14.7 1.6–19.7 0.9–1.6 1.4–2.6 0.8–2.1 0.9–2.4 1.2–4.3 0.7–1.7 0.9–3.1 0.7–1.5

1.9 2.0 2.3b 0.9b 1.4 1.2 1.1 1.9 1.7 2.3 1.2

1.3–2.9 1.3–2.9 2.3–3.6b 0.7–1.0b 1.0–2.3 0.9–1.4 0.8–1.6 1.4–3.5 1.1–2.7 1.5–4.2 0.8–1.9

b

b

IQR: interquartile range; CI: confidence interval; NSCLC: non–small cell lung cancer; SCCA: squamous cell carcinoma antigen; SCC: squamous cell carcinoma; SCLC: small cell lung cancer; CKD: chronic kidney disease. a Excluding CKD 4 and 5 (excluding renal). b Sample size too small for a meaningful comparison.

Figure 2. SCCA for the differential diagnosis of cervical cancer. (a) ROC analysis of SCCA levels as a diagnostic biomarker for differentiating cervical SCC from other patient cohorts. (b) SCCA levels in healthy females and patients with benign gynecological disease, other cervical cancers, SCC cervical cancer, stages I–IIA and IIV–1V. Other cervical cancer group includes patients with adenocarcinoma (n = 12) and other cervical cancers (n = 18). Females only from the apparently healthy individual group (n = 159). For breakdown by stages I–IIA and IIB–IV, n = 1 missing patient is not displayed as staging information was not available. AUC: area under the curve; benign gyn: benign gynecological; SCC: squamous cell carcinoma; SCCA: squamous cell carcinoma antigen.

8

Tumor Biology

Table 4. ROC analysis AUC, sensitivity and specificity results. AUC, % (95% CI) Cervical SCC vs apparently healthy SCC vs benign cervicala SCC vs other cervical cancers Lung SCC vs apparently healthy SCC vs benign lunga SCC vs other NSCLCb Head and neck SCC vs apparently healthy SCC vs benign head and necka SCC vs other head and neck

86.2 (81.8–90.6) 86.3 (81.2–91.3) 78.9 (70.8–87.1) 72.5 (67.8–77.2) 73.8 (68.6–79.1) 71.7 (67.2–76.2) 74.2 (69.2–79.1) 67.5 (59.5–75.6) c

Sensitivity, % (95% CI) At cut-off 2.9 ng/mL 61.4 (52.4–69.9) 61.4 (52.4–69.9) 61.4 (52.4–69.9) At cut-off 3.4 ng/mL 28.4 (22.5–34.9) 28.4 (22.5–34.9) 28.4 (22.5–34.9) At cut-off 2.8 ng/mL 26.6 (19.8–34.3) 26.6 (19.8–34.3) c

Specificity, % (95% CI)

Sensitivity at 95% specificity, % (95% CI)

95.6 (91.1–98.2) 95.0 (86.1–99.0) 86.7 (69.3–96.2)

61.4 (52.4–69.9) 61.4 (52.4–69.9) 55.9 (46.8–64.7)

98.4 (95.9–99.6) 96.7 (91.9–99.1) 91.4 (87.5–94.4)

34.4 (28.1–41.2) 39.1 (32.5–45.9) 17.7 (12.8–23.4)

95.5 (92.1–97.7) 100.0 (91.0–100.0)

26.6 (19.8–34.3) 27.9 (21.0–35.7)

c

c

AUC: area under the curve; CI: confidence interval; ROC: receiver operating characteristic; NSCLC: non–small cell lung cancer; SCC: squamous cell carcinoma; NOS: not otherwise specified. a Benign diseases were lung, head and neck, gynecological. b Excluding NOS/mix. c Sample size too small for a meaningful analysis.

Figure 3. SCCA for the differential diagnosis of lung cancer. (a) SCCA levels in lung SCC versus apparently healthy patients, benign diseases, and other types of lung cancer. (b) ROC analysis of SCCA levels as a diagnostic biomarker for differentiating lung SCC from other patient cohorts. (c) SCCA levels in lung SCC UICC stages. a

Other NSCLC group includes adenocarcinoma (n = 261) + NSCLC other, excluding NSCLC mix/not otherwise specified (n = 18). AUC: area under the curve; NOS: not otherwise specified; NSCLC: non-small cell lung cancer; ROC: receiver operating characteristic; SCC: squamous cell carcinoma; SCCA: squamous cell carcinoma antigen; UICC: International Union Against Cancer.

was applied for each indication. Thus, using the same SCCA cut-off of 2.9 ng/mL, the sensitivity for detecting cervical SCC versus benign disease was 61.4% and the specificity was 95.0%. SCCA levels of 2.9 ng/mL could discriminate between cervical SCC and other cervical cancers with 61.4% sensitivity and 86.7% specificity. Of note is the fact that significantly higher median SCCA levels were observed in the more advanced UICC stages of cervical cancer (IIB–IV) compared with the earlier stages (I–IIA; Figure 2(b)).

Differential diagnosis: lung cancer. Median SCCA levels were slightly but non-significantly higher in the NSCLC

SCC cohort (1.9 (IQR: 0.9–2.4) ng/mL) than in the healthy individuals (1.2 (IQR: 0.8–1.6) ng/mL), other NSCLC patients (1.0 (IQR: 0.7–1.7) ng/mL), and those with benign lung diseases (1.1 (IQR: 0.8–1.6) ng/mL) (Figure 3(a) and Table 3). ROC analysis showed that SCCA was able to differentiate between lung cancer and apparently healthy individuals with an AUC of 72.5% (95% CI: 67.8– 77.2), benign lung diseases with an AUC of 73.8% (95% CI: 68.6–79.1), and patients with NSCLC SCC and those with other NSCLC types (excluding mix/not otherwise specified (NOS)) with an AUC of 71.7% (95% CI: 67.2–76.2; Figure 3(b)).

Holdenrieder et al. The sensitivities at 95% specificity for the comparisons between lung SCC and healthy individuals was 34.4%, benign patients was 39.1%, and other NSCLC was 17.7% (Table 4). The relevant SCCA cut-off based on the 95% specificity for all lung differential diagnosis comparisons was calculated as 3.4 ng/mL. Using this SCCA cut-off of 3.4 ng/mL, it was possible to discriminate between NSCLC SCC and the apparently healthy patient cohort with a sensitivity and specificity of 28.4% and 98.4%, respectively. Thus, 98.4% of apparently healthy individuals had SCCA levels  3.4 ng/mL, while the SCCA level in 28.4% of patients with NSCLC SCC was .3.4 ng/mL. Again, the cut-off of 3.4 ng/mL produced .95% specificity for the comparison with benign lung cancer and other NSCLC, and thus one consistent cutoff level was deemed appropriate. The same SCCA cutoff was able to differentiate between the NSCLC SCC and benign lung diseases groups with a sensitivity and specificity of 28.4% and 96.7%, respectively, and between the NSCLC SCC and other NSCLC patients with a sensitivity of 28.4% and a specificity of 91.4%, respectively. Higher SCCA levels were observed in the more advanced stages of NSCLC SCC, particularly for stages III–IV (2.2 (IQR: 1.3–5.1) ng/mL; p \ 0.001) (Figure 3(c)).

Differential diagnosis: head and neck cancer. Median SCCA levels were slightly higher in the head and neck SCC patient cohort (2.0 (IQR: 1.3–2.9) ng/mL) compared with the apparently healthy individuals (1.2 (IQR: 0.8–1.6) ng/mL) (differences not significant) and similar to those with benign head and neck diseases (1.9 (IQR: 1.4–3.5) ng/mL; Table 3). However, the difference between the median SCCA levels in head and neck SCC versus controls was not as apparent as in the other two malignancies (i.e. NSCLC SCC and cervical SCC). SCCA levels (2.0 ng/mL) were higher in the head and neck SCC cohort than in those with head and neck cancers with other histologies (0.9 ng/mL). ROC analysis showed the AUC for differentiating head and neck SCC versus apparently healthy individuals was 74.2% (95% CI: 69.2–79.1). Head and neck SCC could be distinguished from benign diseases with an AUC of 67.5% (95% CI: 59.5–75.6). The number of patients in the non-SCC head and neck cancer group was too low for a meaningful ROC analysis and it would be difficult to interpret the results. The sensitivities at 95% specificity for the comparisons between head and neck SCC and healthy individuals was 26.6%, benign head and neck patients was 27.9 (Table 4).

9 With an SCCA differential cut-off of .2.8 ng/mL, the sensitivity for head and neck SCC versus apparently healthy patients was 26.6% and the specificity was 95.5%. Using the same cut-off, the sensitivity for detecting head and neck SCC versus benign head and neck diseases was 26.6% (95% CI: 19.8–34.3), with a specificity of 100.0%. No clear correlation was found between SCCA levels and UICC stages in the head and neck malignancy group (data not shown).

Discussion This multi-center study evaluated the technical performance of a new SCC assay for measuring SCCA1 and SCCA2 serum levels in an equimolar manner, the first assay available to do so. Currently, studies on the usefulness of the SCCA biomarker as a tool for the differential diagnosis of patients with SCC are limited. Therefore, the clinical utility of the biomarker for differential diagnosis of cervical, lung, and head and neck SCC was also evaluated. Over 2000 samples were tested from patients from various cohorts, one of the largest studies to do so. Overall, the Elecsys SCC assay showed good precision and agreement with other commercially available assays.17–20 Outliers were removed from the precision results; these were likely due to skin or saliva contamination issues, as is common with SCCA.21 Removal of three outliers was in line with the CLSI EP05-A3 outlier analysis method.38 Even when outliers were included, the technical performance results of the Elecsys SCC assay were well correlated with those of the Architect assay, the most commonly used method. This similar technical performance was despite the antibody differences in the ability to detect both isoforms SCCA1 and SCCA2. It should be noted that the slope of 1.5, in comparison with the Kryptor method, must be taken into consideration when laboratories change methods. The Elecsys SCC ECLIA is the only assay to detect SCCA1 and SCCA2 in equimolar manner; this could be the reason for the differences seen here. In addition, individual differences may occur when using different methods. Therefore, marker kinetics can be interpreted only when the same methods are used for serial determinations. Both European and Chinese samples from healthy patients were analyzed to determine accurate reference ranges for both a Caucasian population and a Chinese one. The median reference range values were 1.1 ng/mL (95th percentile, 2.3 ng/mL) for the European cohort and 1.1 ng/mL (95th percentile, 2.7 ng/mL) for the Chinese cohort. These were both similar to estimations from previous studies: the Architect found 2.1 ng/mL for the 95th percentile in a population of patients in the

10 United States20 or slightly higher than other reported values.13 SCCA levels were far higher in the cervical SCC patient population than the other populations, as has been seen previously.9,10,13 The SCCA levels were not as high in the lung or head and neck SCC cohorts as previously observed.11–13 Gender, age, ethnicity, and smoking status had no significant effect on SCCA levels. This was also true for the apparently healthy patient cohort. In terms of using SCCA as a differential diagnosis tool, the strongest results were observed with the cervical SCC cohort. At 95% sensitivity, the specificity was 61.4% for differentiating healthy and benign cohorts and 55.9% for other cervical malignancies. It was considered useful to have one clinically relevant cut-off to distinguish patients with cervical SCC from other patients: to achieve a 95.6% specificity of SCCA in diagnosing cervical SCC, a cut-off of 2.9 ng/mL was applied. Therefore, in approximately 96% of patients, SCCA levels alone would be able to correctly differentiate between those with cervical SCC versus apparently healthy individuals or those with benign diseases. In 86.6% of patients, an SCCA of  2.9 ng/mL was able to discriminate cervical SCC from other types of cervical cancer. The most recent National Association of Clinical Biochemistry guidelines, published in 2010, state that diagnosis of cervical SCC in all cases should be determined by histopathological findings only; however, the results from our study show that SCCA could also be a sensitive tool for the differential diagnosis of cervical SCC.13 Higher SCCA levels were observed in the more advanced stages of cervical SCC, which is consistent with previous findings.13 The large increase in SCCA levels from stages I–IIA to stages IIB–IV cervical cancer is clinically relevant and should be further exploited for early cancer detection. This large increase between stages IIA and IIB was also reported in previous studies.9,10,13–16,30,49 This increase between IIA and IIB is interesting and, based on these results, using a simple blood test SCCA levels could potentially be a differentiator between these two stages. Historically, stage IIA is still considered ‘‘early stage’’ and upfront surgery is an option.50 However, stage IIB and higher fall into the ‘‘locally advanced’’ category and primary treatment is definitive chemo-radiation. In addition, cervical cancer is highly prevalent in low-income countries with limited resources, for example, for exploratory laparoscopy or advanced imaging and hence SCCA levels could be a useful and cheap quick additional diagnostic test to stratify patients and better allocate limited resources. Along the patient flow, one could envision a continuum for SCCA use. In patients with indeterminate clinical gynecologic findings, elevated SCCA levels would increase suspicion for cervical carcinoma and prevent

Tumor Biology further delay in work-up. Since approximately 10% of cervical carcinomas are not the SCC type, SCCA levels could be used to identify patients with these unclassified cervical malignancies. In addition, in patients with diagnosed cervical SCC, higher SCCA levels would raise the suspicion of more advanced disease, and thus additional tests for staging might be considered, for example, a bone scan to rule out bone metastases. In lung cancer, the ability of SCCA to discriminate between SCC and apparently healthy, benign, and SCLC types was moderate. It is interesting to note that abnormal SCCA levels are suggestive, with high probability, of NSCLC and especially SCC.51,52 In lung cancer, SCC is therefore likely to be most useful when used in conjunction with other biomarkers, such as carcinoembryonic antigen (CEA), cytokeratin-19 fragment (CYFRA 21-1), neuron-specific enolase (NSE), or progastrin-releasing peptide (ProGRP).51,53 Further work to investigate its use with other multi-markers is currently being done. Previous studies have shown SCCA levels to be elevated in patients with head and neck cancer,36,54 and higher in patients with more advanced disease.12,35 However, we found little difference in the SCCA levels of head and neck SCC patients compared with those with benign diseases. In addition, no clear pattern was observed among the TNM staging. The ability of SCCA to discriminate head and neck cancers from apparently healthy patients or those with benign diseases was low, with 26.6% and 27.9% sensitivity, respectively, at 95% specificity. These results suggest that SCCA as a single marker is not useful in the staging or differential diagnosis of head and neck cancer. Potential limitations of the study include the use of samples from serum banks and employing precautions to minimize the risk of possible contamination, which may not be wholly representative of routine working in clinical laboratories. In conclusion, this multi-center study showed that the Elecsys SCC assay had good technical performance for the measurement of SCCA in clinical practice and was comparable with the two other commercially available assay methods. To our knowledge, this is one of the largest studies into SCCA for the differential diagnosis of cervical, lung, and head and neck SCC, including controls from two continents. The result showed that SCCA levels may be useful for the differential diagnosis of SCC types, particularly in patients with cervical cancer, and could be a simple tool for diagnosing patients with cervical SCC. Acknowledgements The authors would like to thank Wolfgang Junge, Laboratorium fu¨r Klinische Forschung GmbH, Kiel, Germany, for his contribution to the reference range sample

Holdenrieder et al. collection, and to Jeroen de Jong, The Netherlands Cancer Institute, Amsterdam, The Netherlands, for his clinical input. Birgit Wehnl, Roche Diagnostics GmbH, Penzberg, Germany, provided study management support and advice. Study monitoring was performed by Monika Steclik (Triga-S e. K., Habach, Germany) and Silvia Barbara Schmid (Roche Diagnostics GmbH, Penzberg, Germany). Christina Rabe and Katharina Buck, Roche Diagnostics GmbH, Penzberg, Germany, provided statistical support and advice. Scientific consulting was provided by Marcus-Rene Lisy (Roche Diagnostics GmbH, Penzberg, Germany). Medical writing assistance for this manuscript was provided by Kim Brown (Roche Diagnostics International Ltd, Rotkreuz, Switzerland). Stefan Holdenrieder, Rafael Molina, Ling Qiu, Xiuyi Zhi, Sandra Rutz, and Catharina M Korse were involved in data collection, data interpretation, manuscript writing, and approved the manuscript for submission. Christine Engel, Pia Kasper-Sauer, and Farshid Dayyani were involved in data interpretation, manuscript writing, and approved the manuscript for submission.

Compliance with ethical standards All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all individual participants included in the study.

Declaration of conflicting interests The author declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Sandra Rutz, Christine Engel, and Pia Kasper-Sauer are employees of Roche Diagnostics. Farshid Dayyani was an employee of Roche Diagnostics at the time of the study. Stefan Holdenrieder, Rafael Molina, and Catharina M Korse have received grants/research support from Roche Diagnostics.

Funding The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The study was funded by Roche Diagnostics. Medical writing support was provided by Emma McConnell of Gardiner-Caldwell Communications. This study was sponsored by Roche Diagnostics.

References 1. Yan W, Wistuba II, Emmert-Buck MR, et al. Squamous cell carcinoma—similarities and differences among anatomical sites. Am J Cancer Res 2011; 1: 275–300. 2. Spiro SG, Tanner NT, Silvestri GA, et al. Lung cancer: progress in diagnosis, staging and therapy. Respirology 2010; 15: 44–50. 3. Broaddus VC, Mason RJ, Ernst JD, et al. Murray & Nadel’s textbook of respiratory medicine. 6th ed. Philadelphia, PA: Saunders, 2016.

11 4. Marur S and Forastiere AA. Head and neck cancer: changing epidemiology, diagnosis, and treatment. Mayo Clin Proc 2008; 83: 489–501. 5. Sheets E and Goodman H. The cervix. In: Berkowitz R and Barbieri R (eds) Kistner’s gynecology principles and practice. St. Louis, MI: MD Consult, LLC, 1995, pp. 103–136. 6. Kato H and Torigoe T. Radioimmunoassay for tumor antigen of human cervical squamous cell carcinoma. Cancer 1977; 40: 1621–1628. 7. Cataltepe S, Gornstein ER, Schick C, et al. Co-expression of the squamous cell carcinoma antigens 1 and 2 in normal adult human tissues and squamous cell carcinomas. J Histochem Cytochem 2000; 48: 113–122. 8. Roijer E, de Bruijn HW, Dahlen U, et al. Squamous cell carcinoma antigen isoforms in serum from cervical cancer patients. Tumour Biol 2006; 27: 142–152. 9. Gadducci A, Tana R, Cosio S, et al. The serum assay of tumour markers in the prognostic evaluation, treatment monitoring and follow-up of patients with cervical cancer: a review of the literature. Crit Rev Oncol Hematol 2008; 66: 10–20. 10. Yoon SM, Shin KH, Kim JY, et al. The clinical values of squamous cell carcinoma antigen and carcinoembryonic antigen in patients with cervical cancer treated with concurrent chemoradiotherapy. Int J Gynecol Cancer 2007; 17: 872–878. 11. Molina R, Filella X, Auge JM, et al. Tumor markers (CEA, CA 125, CYFRA 21-1, SCC and NSE) in patients with non-small cell lung cancer as an aid in histological diagnosis and prognosis. Comparison with the main clinical and pathological prognostic factors. Tumour Biol 2003; 24: 209–218. 12. Molina R, Filella X, Torres MD, et al. SCC antigen measured in malignant and nonmalignant diseases. Clin Chem 1990; 36: 251–254. 13. Sturgeon CM, Duffy MJ, Hofmann BR, et al. National Academy of Clinical Biochemistry Laboratory Medicine Practice Guidelines for use of tumor markers in liver, bladder, cervical, and gastric cancers. Clin Chem 2010; 56: e1–e48. 14. Schmidt-Rhode P, Schulz KD, Sturm G, et al. Squamous cell carcinoma antigen for monitoring cervical cancer. Int J Biol Markers 1988; 3: 87–94. 15. Gaarenstroom KN, Bonfrer JM, Kenter GG, et al. Clinical value of pretreatment serum Cyfra 21-1, tissue polypeptide antigen, and squamous cell carcinoma antigen levels in patients with cervical cancer. Cancer 1995; 76: 807–813. 16. Gaarenstroom KN, Kenter GG, Bonfrer JM, et al. Can initial serum cyfra 21-1, SCC antigen, and TPA levels in squamous cell cervical cancer predict lymph node metastases or prognosis? Gynecol Oncol 2000; 77: 164–170. 17. Sandri MT, Salvatici M, Mauro C, et al. Detection of squamous cell carcinoma antigen with two systems in the follow-up of patients with cervical cancer. Int J Biol Markers 2013; 28: 313–317. 18. Serdarevic N and Stanciu AE. Comparison of Architect i2000 for determination of SCC with IMX determination of SCC with different methods. Clin Lab 2013; 59: 1129–1133.

12 19. Dipalo M, Gnocchi C, Aloe R, et al. Analytical assessment of the novel Maglumi squamous cell carcinoma antigen (SCCA) immunoluminometric assay. Ann Transl Med 2015; 3: 351. 20. Erickson JA, Lu J, Smith JJ, et al. Immunoassay for quantifying squamous cell carcinoma antigen in serum. Clin Chem 2010; 56: 1496–1499. 21. Torre GC. SCC antigen in malignant and nonmalignant squamous lesions. Tumour Biol 1998; 19: 517–526. 22. Campbell B and De’Ambrosis B. Squamous cell carcinoma antigen in patients with cutaneous disorders. J Am Acad Dermatol 1990; 22: 639–642. 23. Cases A, Filella X, Molina R, et al. Tumor markers in chronic renal failure and hemodialysis patients. Nephron 1991; 57: 183–186. 24. Trape J, Filella X, Alsina Donadeu-M, et al. Increased plasma concentrations of tumour markers in the absence of neoplasia. Clin Chem Lab Med 2011; 49: 1605–1620. 25. Kahn N, Riedlinger J, Roessler M, et al. Blood-sampling collection prior to surgery may have a significant influence upon biomarker concentrations measured. Clin Proteomics 2015; 12: 19. 26. Li X, Zhou J, Huang K, et al. The predictive value of serum squamous cell carcinoma antigen in patients with cervical cancer who receive neoadjuvant chemotherapy followed by radical surgery: a single-institute study. PLoS ONE 2015; 10: e0122361. 27. Esajas MD, Duk JM, de Bruijn HW, et al. Clinical value of routine serum squamous cell carcinoma antigen in follow-up of patients with early-stage cervical cancer. J Clin Oncol 2001; 19: 3960–3966. 28. Hirakawa M, Nagai Y, Inamine M, et al. Predictive factor of distant recurrence in locally advanced squamous cell carcinoma of the cervix treated with concurrent chemoradiotherapy. Gynecol Oncol 2008; 108: 126–129. 29. Micke O, Prott FJ, Schafer U, et al. The impact of squamous cell carcinoma (SCC) antigen in the follow-up after radiotherapy in patients with cervical cancer. Anticancer Res 2000; 20: 5113–5115. 30. Duk JM, Groenier KH, de Bruijn HW, et al. Pretreatment serum squamous cell carcinoma antigen: a newly identified prognostic factor in early-stage cervical carcinoma. J Clin Oncol 1996; 14: 111–118. 31. Yazigi R, Munoz AK, Richardson B, et al. Correlation of squamous cell carcinoma antigen levels and treatment response in cervical cancer. Gynecol Oncol 1991; 41: 135–138. 32. Diez M, Torres A, Maestro ML, et al. Prediction of survival and recurrence by serum and cytosolic levels of CEA, CA125 and SCC antigens in resectable non-smallcell lung cancer. Br J Cancer 1996; 73: 1248–1254. 33. Sanchez De Cos J, Masa F, de la Cruz JL, et al. Squamous cell carcinoma antigen (SCC Ag) in the diagnosis and prognosis of lung cancer. Chest 1994; 105: 773–776. 34. Vassilakopoulos T, Troupis T, Sotiropoulou C, et al. Diagnostic and prognostic significance of squamous cell carcinoma antigen in non-small cell lung cancer. Lung Cancer 2001; 32: 137–144. 35. Barak V, Meirovitz A, Leibovici V, et al. The diagnostic and prognostic value of tumor markers (CEA, SCC,

Tumor Biology

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

47. 48.

49.

50.

CYFRA 21-1, TPS) in head and neck cancer patients. Anticancer Res 2015; 35: 5519–5524. Imai R, Takenaka Y, Yasui T, et al. Prognostic significance of serum squamous cell carcinoma antigen in patients with head and neck cancer. Acta Otolaryngol 2015; 135: 295–301. Molina R, Filella X, Lejarcegui JA, et al. Prospective evaluation of squamous cell carcinoma and carcinoembryonic antigen as prognostic factors in patients with cervical cancer. Tumour Biol 2003; 24: 156–164. Clinical Laboratory Standards Institute. EP05-A3 evaluation of precision of quantitative measurement methods: approved guideline. 3rd ed. Wayne: Clinical and Laboratory Standards Institute, 2014. Passing H and Bablok W. A new biometrical procedure for testing the equality of measurements from two different analytical methods. Application of linear regression procedures for method comparison studies in clinical chemistry, part I. J Clin Chem Clin Biochem 1983; 21: 709–720. Clinical and Laboratory Standards Institute. EP09-A3 measurement procedure comparison and bias estimation using patient samples; approved guideline. 3rd ed. Wayne: Clinical and Laboratory Standards Institute, 2013. Union for International Cancer Control (UICC). TNM classification of malignant tumours. 7th ed. Hoboken: Wiley-Blackwell, 2009. Tramonti G, Ferdeghini M, Donadio C, et al. Renal function and serum concentration of five tumor markers (TATI, SCC, CYFRA 21-1, TPA, and TPS) in patients without evidence of neoplasia. Cancer Detect Prev 2000; 24: 86–90. Tong HL, Dong ZN, Wen XY, et al. Impact of chronic kidney disease on serum tumor markers concentrations. Chin Med J 2013; 126: 274–279. Matsushita K, Mahmoodi BK, Woodward M, et al. Comparison of risk prediction using the CKD-EPI equation and the MDRD study equation for estimated glomerular filtration rate. JAMA 2012; 307: 1941–1951. Kunst A, Busse Grawitz A, Engeldinger W, et al. WinCAEv—a new program supporting evaluations of reagents and analysers. Clin Chim Acta 2005; 355S: 361. Henny J. The IFCC recommendations for determining reference intervals: strengths and limitations. J Lab Med 2009; 33: 45–51. Hahn GJ and Meeker WG. Statistical intervals: a guide for practitioners. 1st ed. New York: Wiley, 1991. Pina TC, Zapata IT, Lopez JB, et al. Tumor markers in lung cancer: does the method of obtaining the cut-off point and reference population influence diagnostic yield? Clin Biochem 1999; 32: 467–472. Hong JH, Tsai CS, Chang JT, et al. The prognostic significance of pre- and posttreatment SCC levels in patients with squamous cell carcinoma of the cervix treated by radiotherapy. Int J Radiat Oncol Biol Phys 1998; 41: 823–830. NCCN National Comprehensive Cancer Network (NCCN). Cervical cancer (version 1. www.nccn.org/professionals/physician_gls/PDF/cervical.pdf (2017, accessed 31 March 2017).

Holdenrieder et al. 51. Song WA, Liu X, Tian XD, et al. Utility of squamous cell carcinoma antigen, carcinoembryonic antigen, Cyfra 21-1 and neuron specific enolase in lung cancer diagnosis: a prospective study from China. Chin Med J 2011; 124: 3244–3248. 52. Molina R, Bosch X, Auge JM, et al. Utility of serum tumor markers as an aid in the differential diagnosis of patients with clinical suspicion of cancer and in patients with cancer of unknown primary site. Tumour Biol 2012; 33: 463–474.

13 53. Molina R, Marrades RM, Auge JM, et al. Assessment of a combined panel of six serum tumor markers for lung cancer. Am J Respir Crit Care Med 2016; 193: 427–437. 54. Koch T, Eiffert H and Spindler MB. Relevance of the new tumor marker SCC (squamous cell carcinoma antigen) for the diagnosis and follow-up control of squamous epithelial carcinoma of the head and neck. HNO 1989; 37: 454–459.