Reproducibility of Mass Spectrometry Based Protein Profiles for Diagnosis of Breast Cancer across Clinical Studies: A Systematic Review Anne K. Callesen,*,†,‡ Werner Vach,§ Per E. Jørgensen,‡ Søren Cold,| Ole Mogensen,⊥ Torben A. Kruse,‡ Ole N. Jensen,*,† and Jonna S. Madsen‡ Protein Research Group, Department of Biochemistry and Molecular Biology, University of Southern Denmark, Odense, Denmark, Department of Biochemistry, Pharmacology and Genetics, Odense University Hospital, Odense, Denmark, Department of Statistics, University of Southern Denmark, Odense, Denmark, Department of Oncology, Odense University Hospital, Odense, Denmark, and Department of Gynecology and Obstetrics, Odense University Hospital, Odense, Denmark Received November 15, 2007
Serum protein profiling by mass spectrometry has achieved attention as a promising technology in oncoproteomics. We performed a systematic review of published reports on protein profiling as a diagnostic tool for breast cancer. The MEDLINE, EMBASE, and COCHRANE databases were searched for original studies reporting discriminatory protein peaks for breast cancer as either protein identity or as m/z values in the period from January 1995 to October 2006. To address the important aspect of reproducibility of mass spectrometry data across different clinical studies, we compared the published lists of potential discriminatory peaks with those peaks detected in an original MALDI MS protein profiling study performed by our own research group. A total of 20 protein/peptide profiling studies were eligible for inclusion in the systematic review. Only 3 reports included information on protein identity. Although the studies revealed a considerable heterogeneity in relation to experimental design, biological variation, preanalytical conditions, methods of computational data analysis, and analytical reproducibility of profiles, we found that 45% of peaks previously reported to correlate with breast cancer were also detected in our experimental study. Furthermore, 25% of these redetected peaks also showed a significant difference between cases and controls in our study. Thus, despite known problems related to reproducibility, we were able to demonstrate overlap in peaks between clinical studies indicating some convergence toward a set of common discriminating, reproducible peaks for breast cancer. These peaks should be further characterized for identification of the protein identity and validated as biomarkers for breast cancer. Keywords: Protein profiling • Proteomics • Breast cancer • Mass spectrometry, Systematic review
Introduction One in every nine women will develop breast cancer during her lifetime, and therefore, this disease has a major impact on female health. Early detection and intervention are important factors affecting the outcome of the disease. Much attention has been given to the potential use of proteomic technologies for discovery and characterization of clinically valuable biom* Correspondence to Anne K. Callesen or Ole N. Jensen, Protein Research Group, Department of Biochemistry and Molecular Biology, University of Southern Denmark, Campusvej 55, 5230 Odense M, Denmark. Phone: +45 6550 2342. Fax: +45 6550 2467. E-mails: (A.K.C.) anne.callesen@ouh .fyns-amt.dk, (O.N.J.)
[email protected]. † Department of Biochemistry and Molecular Biology, University of Southern Denmark. ‡ Department of Biochemistry, Pharmacology and Genetics, Odense University Hospital. § Department of Statistics, University of Southern Denmark. | Department of Oncology, Odense University Hospital. ⊥ Department of Gynecology and Obstetrics, Odense University Hospital. 10.1021/pr800115f CCC: $40.75
2008 American Chemical Society
arkers with the ability to aid early diagnosis or guide optimal treatment decisions.1 Mass spectrometry (MS) platforms have revolutionized proteomics. MS analysis of polypeptides in biological fluids can potentially provide diagnostic and prognostic information for cancer and has driven the emergence of the research area of clinical proteomics.2 Surface-enhanced laser desorption/ionization mass spectrometry (SELDI MS) and matrix-assisted laser desorption/ionization mass spectrometry (MALDI MS) have been investigated for polypeptide profiling in clinical proteomics. The SELDI MS approach for diagnosis of breast cancer was used in the majority of the studies,3–21 and although these studies have shown potential for MS as an early diagnostic tool, the reproducibility of results remains a challenge. The general usefulness of the SELDI MS technology has been questioned due to poor analytical performance in the early SELDI MS instruments,22 and it has been argued that problems of reproducibility and standardization of SELDI must be addressed before this technique can become a routine Journal of Proteome Research 2008, 7, 1395–1402 1395 Published on Web 02/28/2008
reviews
Callesen et al. 23
clinical diagnostic tool. Therefore, the more robust and flexible method of MALDI MS has received increased attention for clinical proteomics applications.22,24–27 Pre- and postanalytical conditions are important factors for reproducibility of clinical studies. Consequently, the comparison of different studies is difficult due to a number of variable factors relating to experimental design, collection of specimen, biological variation, preanalytical conditions, and mass spectrometry performance, as well as methods of computational data analysis and analytical/technical reproducibility of serum profiles.25,28–31 On the other hand, a demonstration of an overlap in peaks between studies would indicate presence of a set of common discriminating peaks for breast cancer. Until now, no systematic review on the status of MS-based proteomics as a diagnostic tool for breast cancer has been performed. The aim of the current study was to provide such a systematic review including original studies that reported protein profiling data with discriminatory protein peaks for breast cancer detection as either protein identity or as m/z values. To address the important question of reproducibility across clinical studies, we compared the published lists of discriminatory peaks with peaks detected in an experimental study performed by our own research group and published elsewhere in this journal.32
Methods Data Sources and Searches. Data Sources, Search Terms, and Strategies. We conducted the review by using the following protocol. Two reviewers (A.K.C. and J.S.M.) independently searched the MEDLINE, EMBASE, and COCHRANE databases for studies published up to and including October 2006. We developed a strategy for searching MEDLINE, accessed through PubMed, that was based on analysis of the Medical Subject Heading terms and text words of key articles identified a priori. EMBASE was assessed through SCOPUS covering 100% of articles present in EMBASE (and MEDLINE). To ensure inclusion of all relevant studies (high sensitivity in relation to identify relevant studies), the final search-string was broad and thus a priori expected to have a relatively low specificity. The search terms were (proteomic* analysis OR mass spectrometry OR MS OR MALDI OR surface-enhanced laser desorption/ionizationtime-of-flight OR SELDI-TOF) AND (proteome OR tumor marker* OR proteomic* OR biomarker* OR serum biomarker* OR serum profiling) AND (solid tumor AND cancer) OR (breast cancer OR breast cancers). No language restrictions were applied in the retrieval of citations. Inclusion and Exclusion Criteria. Primary proteomic research studies in breast cancer were selected. Only studies reporting the protein identity and/or the m/z value of the detected peaks were included. No restriction was made according to the test material used, and consequently, a wide variety of biological fluids and tissues were included. Cell cultures and 2D-gel studies were excluded, and only studies utilizing MS were included. In addition, nonhuman studies were not included. Data Extraction and Presentation. For each study, information on study design and population, test material, and mass spectrometry technique used was noted. Furthermore, information on protein identity, detection range, and m/z value of the discriminatory protein peaks was included. Comparison with Results Obtained in Our Own Experimental Study. The published discriminating protein peaks in the reviewed studies were compared with results 1396
Journal of Proteome Research • Vol. 7, No. 4, 2008
Figure 1. Study flow diagram.
obtained in our own experimental study.32 Only studies reporting mass accuracy less than 100 Da were included in this comparison to allow sufficiently precise cross-study comparisons. In our experimental study, we could identify an overall profile with 533 peaks visible in at least 10 out of 76 subjects, across the mass range 3-25 kDa. Within this profile, 72 peaks showed a significant difference at the 5% level. A list of the most discriminating protein peaks and a detailed description of the entire experimental and statistical strategy for mass spectrometry based serum protein profiling used in this study is presented elsewhere in this journal.32 Role of the Funding Sources. This work was supported by grants from the Danish Cancer Society, Ministry of the Interior and Health, Lands-landsdelspuljen, and a donation from Elly Jacobsen (to A.K.C). O.N.J. is a Lundbeck Foundation Research Professor. The funding sources had no role in conducting the review or in preparing and submitting the manuscript.
Results The literature search identified 424 articles as potentially relevant. After reviewing titles and abstracts, 311 nonrelevant articles were excluded, leaving a total of 113 articles for fulllength review. In this process, further articles were identified as nonrelevant to this systematic review of reasons stated in Figure 1, which presents the study flow diagram in identifying relevant articles. Finally, 20 articles were left as relevant according to the predefined criteria. These articles are presented in Table 1. As reported in Table 1, only 3 of the 20 studies provided information on protein identity. There was a considerable heterogeneity between the included studies. Major differences in study design, population, sample size, and test material existed. Also the reporting of number and masses of discriminating peaks found in the different mass spectrometry protein profiling studies varied considerable. Thus, one of the 20 studies did not publish a list of identified discriminatory peaks, and four studies reported peak values only with a precision of g100 Da, which were judged insufficient for cross-study comparisons. For the remaining 15 studies, we performed a detailed comparison of the published peaks with those included in our overall profile in an experimental study (Tables 2 and 3).
reviews
MS-Based Protein Profiles for Diagnosis of Breast Cancer a
Table 1. Characteristics of Reviewed Studies
author
Becker et al.4
study design and population
detection range [kDa]
protein ID (yes/no)
NR
No
SELDI MS (gold chip) Serum SELDI MS (SAX; WCX; IMAC-cu) Case/control study (49 BC, 33 H, 51 BBD) Serum SELDI MS ( H4) Comparison of cancerous and noncancerous NAF, DLF SELDI MS breasts (23 ULPIC) (WCX; IMAC) Comparison of pre and postsurgery (16 pre and Serum SELDI MS postsurgery, 15 H) ( SAX, IMAC-cu) Comparison of BRCA-1 cancers and BRCA-1 carriers (15 BRCA-1 cancer, 15 BRCA-1 carriers) Comparison of lymph node status (27 LN pos, 71 LN neg) Case/control study (103 BC, 41 H, 25 BBD) Serum SELDI MS (IMAC-Ni) Validation of earlier published biomarkers (32 Serum SELDI MS DCIS, 61 LIBC, 37 BBD, 46 H) (IMAC-Ni) Biomarker discovery and comparison of NAF, DLF SELDI MS cancerous and noncancerous breasts (NAF: 5 (IMAC-Cu) ULPIC, 5 H (training sample)) (DLF: 42 ULPIC (25 cancerous breasts, 17 noncancerous breasts), 42 HRW) Validation of biomarkers published by Li et al.3 Serum SELDI MS (49 BC, 27 H, 13 BBD) (IMAC-Ni) Comparison of cancerous and noncancerous DLF SELDI MS breasts (16 ULBC - unilateral breast carcinoma) (SAX)
1–3
No
NR
No
2–30
No
7–30
No
1.5–20
No
Paweletz et al.14
Case/control study (12 BC, 15 H)
NAF
Pawlik et al.15
Comparison of cancerous and noncancerous breasts (23 ULPIC, 5 H)
NAF, DLF
Pusztai et al.16
Response to chemotherapy treatment (69 BC, 15 H) Case/control study (20 BC, 13 H)
Plasma
Sauter et al.18
Biomarker discovery (81 BBD, 6 ADH, 5 DCIS, 22 IBC)
NAF, DLF
Streckfus et al.19
Case/control study (3 DCIS, 3 H)
Saliva
Traub et al.20
Biomarker discovery (20 tumors)
Tissue
Villanueva et al.22
Comparison of different cancer types (32 prostate cancer, 21 BC, 20 bladder, 33 H) Case/control study (45 BC, 47 H, 42 BBD)
Serum
6
Heike et al. Hu et al.7
Kuerer et al.8 Laronga et al.9
3
Li et al.
Li et al.10 Li et al.11
Mathelin et al.12 Mendrinos et al.13
17
Sauter et al.
Vlahou et al.21
Serum
mass spectrometry
SELDI MS (IMAC-Cu)
Caputo et al.5
Comparison of BRCA-1 cancers and BRCA-1 carriers (15 BRCA-1 cancer, 15 BRCA-1 carriers, 16 SBC, 16 H) Comparison of different cancer types (10 melanoma, 10 BC, 10 H) Comparison of pre and post-treatment (6 docetaxal infusion treatment)
test material
Plasma
NAF, DLF
Serum
m/z value [Da]
List of 37 peaks (published) (BRCA-1 cancer vs BRCA-1 carriers) 2236.1; 2356.3 (BC vs H and melanoma) 7790; 9285 (pre vs post-treatment) 5.7 × 103; 8.9 × 103; 17.3 × 103; 26.2 × 103 (BC vs H) Total of 463 peaks (not published) (cancerous vs noncancerous breasts) 2146; 3161; 3686; 3820; 6679 (pre and post vs H) 107 peaks (not published) 6 peaks (not published)
NR
No
2–150
Yes
3–135
Yes
3–10
No
NR
No
SELDI MS (C16)
NR
No
SELDI MS (WCX; IMAC-Cu) SELDI MS (IMAC-Cu) SELDI MS (H4; NP; SAX) SELDI MS (H4; NP; SAX) SELDI MS (H4; SAX; WCX) SELDI MS (SAX; WCX; H4; IMAC-Cu) MALDI-TOF MS SELDI MS (SAX; IMAC-Cu)
NR
No
NR
No
5–40
No
NR
No
NR
No
NR
No
0.7–15
Yes
NR
No
4.3 × 103; 8.1 × 103; 8.9 × 103(BC vs H + BBD) 8116; 8926 (DCIS + LIBC vs BBD + H) 3375; 3447; 3490; 4079; 4680 (cancerous vs noncancerous breasts) 4286; 4302; 8919; 8961 (BC vs H + BBD) (30.4 × 103/31 × 103) (cancerous vs noncancerous breasts) 4233.09; 9470.0 (BC vs H up regulated in BC) 3415 .6; 4149.7 (BC vs H - up regulated in H) List of 27 peaks (published) (H vs ULPIC) 3165; 3440; 4115; 4444; 8940 (H vs BC) 6500; 8000; 15940; 28100; 31770 (BC vs H) 5200; 11880; 13880; 33400 (BBD + ADH vs DCIS +IBC) 18 × 103; 113 × 103; 170 × 103; 228 × 103; 287 × 103 (DCIS vs H) List of 72 peaks (published) List of 25 peaks (published) (H vs BC) 2.95 × 103; 3.68 × 103; 3.94 × 103; 3.97 × 103; 4.27 × 103 (BC vs H) 6.43 × 103; 7.48 × 103; 8.61 × 103(BC vs BBD)
a Abbreviations: SBC, sporadic breast cancer; H, healthy; BC, breast cancer; BBD, benign breast disease; NAF, nipple aspirate fluid; DLF, ductal lavage fluid; ULPIC, unilateral primary invasive cancer (stage I and II); LN, lymph node; DCIS, ductal carcinoma in situ; LIBC, locally invasive breast cancer; HRW, high-risk women (multiple ducts repeated lavage); ADH, atypical ductal hyperplasia; IBC, invasive breast cancer; NR, not reported.
A total of 207 peaks have been published recently. Actually, 10 of the previously published peaks have been published in more than one study: 1 peak (m/z 4302.3) was found in 4 studies,3,4,12,20 1 peak (m/z 2953.3) was found in 3 studies,4,15,21 and finally 8 peaks were found in 2 studies. In Table 2, we can observe that about 45% of the previously published discriminating peaks were also represented in the
overall protein profile obtained in our experimental study. About 25% of the previously published discriminatory peaks that were also present in our protein profile were furthermore significant at the 5% level in our study. In addition, the fraction of peaks that were determined as significant in our study was significantly above chance level in 6 of the 15 studies presented in Table 2. The frequency of a significant finding in our study Journal of Proteome Research • Vol. 7, No. 4, 2008 1397
reviews
Callesen et al. 32 a
Table 2. Reviewed Studies Reporting Discriminating Protein Peaks Compared to Results from an Original Study published peaks in reviewed studies
corresponding peaks identified in original study32 properties of corresponding peaks
test material
study
NAF Li et al.11 NAF Pawlik et al.15 NAF Paweletz et al.14 NAF Sauter et al.17 NAF Sauter et al.18 Plasma Caputo et al.5 Plasma Pusztai et al.16 Serum Becker et al.4 Serum Heike et al.6 Serum Laronga et al.9 Serum Li et al.10 Serum Mathelin et al.12 Serum Villanueva et al.22 Serum Vlahou et al.21 Tissue Traub et al.20 All studies in total
number of published peaks
corresponding peaks in our protein profile
5 27 4 5 4 2 5 37 2 5 2 4 25 8 72 207
3 15 3 0 0 2 1 20 0 3 2 1 18 5 20 93
p < 0.05 in our study
additional same tendency in comparing cases and controls
difference of m/z values (published study-our study) [Da]
n
%
n
%
mean
2 6 1 0 1 6 0 0 0 1 2 4 23
67* 40* 33 0 100* 30* 0 0 0 6 40* 20* 25*
2 5 1 1 2 0 ? 3
100 83 100 100 33 0 ? 75
-2.17 3.71* 1.33 1.65 -2.20 0.83 2.03 -4.1 -0.30 -1.14 ** 0.86
range
[-1.5;-3.1] [-5.3;5.4] [-4;1.8] [0.3;3.0] [-2.2] [-6.7;8.1] [1.1;3.7] [-2.3;-5.9] [-0.3] [-3.5;4.2] ** [-4.8;6.9]
a Abbreviation: NAF, nipple aspirate fluid. Correspondence is based on (2‰ of the m/z value in the mass range up to 2.5 kDa, and 1‰ of the m/z value in the mass range up to 25 kDa. (*) Significantly larger than 5% at the 5% level. (**) Not calculated due to mass accuracy >10 Da. NOTE: Only studies with a reported mass accuracy less than 100 Da are included.
increases from 21% for peaks only published once to 38% and 50% in peaks published twice or three/four times. However, this trend was not significant (p ) 0.19). For 5 out of 7 studies, a majority of the previously published peaks show the same tendency in our study as in the original study when comparing cases and controls (Table 3). Furthermore, slightly systematic differences in the peak location between the studies were observed as seen in Table 3. In studies with few corresponding peaks, often all differences were positive or negative, whereas in studies with several corresponding peaks, the mean difference was often significantly different from zero. These differences suggest that the current calibration procedures are not perfect, but that the differences are small enough to allow cross study comparisons of reproducibility.
Discussion This study provides the first systematic review regarding the current status of MS-based protein profiling as a diagnostic tool for breast cancer. A total of 20 MS studies were included that reported discriminatory protein peaks as protein identification or as m/z values for breast cancer detection (Table 1). Only 3 studies reported on protein identification based on discriminatory peaks. It has been difficult to obtain reproducible results among different laboratories,28,33 and the lack of reproducibility of MS protein profiling has been discussed as a major problem. This important problem clearly appears from the presentation of primary research results in diagnostic proteomics studies as presented in Table 1, where only a minority of previously published peaks could be reproduced by other research groups. About 100 peaks from the overall protein profile used in our study have been mentioned as potential markers in previously published proteomic studies (Table 2), and only 25% of these markers could be “validated”, that is, showed a difference between cases and controls at the 5% level in our study. On 1398
Journal of Proteome Research • Vol. 7, No. 4, 2008
the other hand, this degree of reproducibility is clearly above chance level. This is also corroborated by the fact that the fraction of validated peaks was significantly above chance level in 6 of the 12 studies with at least one published peak covered by our overall profile. The current systematic review on proteomics in breast cancer reflected a considerable heterogeneity between the included studies. In several studies, the sample size was small and provided limited clinical details, which makes it difficult to evaluate the clinical relevance. In addition, the reporting of preanalytical factors as sample collection, processing, and storage was, in many cases, sparse which may be a problem as it is increasingly being recognized that preanalytical effect/ factors can exert marked influences on the results. The reporting of number and masses of discriminating peaks found in the different studies also varied considerably. In addition, in several studies, the reported mass values were not presented with sufficient precision and thus hampered comparative analysis.5,13 This underlines the importance of a comprehensive reporting of both preanalytical factors and methodological aspect in such diagnostic studies to allow comparison between studies and in relation to performing systematic reviews.34 It is important to note that despite a wide diversity among the presented studies with regard to sample size, biological material, mass spectrometer, and study design, we found reproducibility above chance level using the results of our own experimental and statistical study as a reference. No clear relation could be found between the material used in the different studies and the degree of reproducibility of peaks. Moreover, the overlap with the only study using MALDI MS22 (and the same test material) was actually poor. Limitations of the Review. Publication bias is possible, and although a comprehensive and systematic search strategy of the literature was performed, relevant studies may have been missed.
reviews
MS-Based Protein Profiles for Diagnosis of Breast Cancer 32 a
Table 3. Comparison of Reviewed Studies with an Original Study
results in original study32
results in published study study
Becker et al.4 Becker et al.4 Becker et al.4 Becker et al.4 Becker et al.4 Becker et al.4 Becker et al.4 Becker et al.4 Becker et al.4 Becker et al.4 Becker et al.4 Becker et al.4 Becker et al.4 Becker et al.4 Becker et al.4 Becker et al.4 Becker et al.4 Becker et al.4 Becker4 Becker et al.4 Caputo et al.5 Caputo et al.5 Laronga et al.9 Laronga et al.9 Laronga et al.9 Li et al.10 Li et al.10 Li et al.11 Li et al.11 Li et al.11 Mathelin et al.12 Paweletz et al.14 Paweletz et al.14 Paweletz et al.14 Pawlik et al.15 Pawlik et al.15 Pawlik et al.15 Pawlik et al.15 Pawlik et al.15 Pawlik et al.15 Pawlik et al.15 Pawlik et al.15 Pawlik et al.15 Pawlik et al.15 Pawlik et al.15 Pawlik et al.15 Pawlik et al.15 Pawlik et al.15 Pawlik et al.15 Pusztai et al.16 Traub et al.20 Traub et al.20 Traub et al.20 Traub et al.20 Traub et al.20 Traub et al.20 Traub et al.20 Traub et al.20 Traub20 Traub et al.20 Traub et al.20 Traub et al.20 Traub et al.20 Traub et al.20
m/z value [M]
m/z value [M + H]+
p-value
tendency
mean m/z value [M + H]+
p-value
tendency
1588.673 2956.088 3278.707 3779.663 3975.987 4306.0 4493.134 4648.0 5341.586 5909.051 6539.982 6807.266 7057.232 7191.317 7347.652 7569.754 7633.829 7984.951 8138.558 8444.507 2236.1 2356.3 3161 3686 6679 8116 8926 3375 3447 3490 4302 4149.7 4233.09 9470.0 952.59 2310.84 2791.86 2956.86 3017.85 3284.74 4182.65 4205.39 5092.29 5101.80 6307.68 6721.72 8385.73 8419.34 8573.09 4115 1168 3447 4183 4190 4305 4674 4821 4832 5096 5728 6362 6899 6917 7009
0.020278 0.005445 1.05 × 10-5 1.81 × 10-5 3.06 × 10-5 0.034501 0.029757 0.004128 9.18 × 10-7 1.02 × 10-7 0.007129 0.001905 0.002211 0.000145 0.008498 0.028663 0.000879 0.007786 7.89 × 10-7 1.33 × 10-6
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
1591.4 2953.3 3281.5 3783.0 3977.4 4302.3 4497.5 4644.1 5336.0 5903.3 6541.8 6801.9 7063.9 7189.1 7344.0 7561.7 7632.4 7982.1 8138.3 8450.6 2233.1 2356.0 3159.9 3682.3 6677.7 8121.9 8928.3 3378.1 3448.9 3491.5 4302.3 4153.7 4231.3 9471.2 951.3 2307.0 2790.8 2953.3 3014.7 3281.5 4178.5 4210.2 5087.0 5099.4 6302.3 6726.8 8388.8 8414.5 8578.4 4117.2 1168.1 3448.9 4178.5 4190.1 4302.3 4675.2 4819.2 4831.0 5099.4 5729.9 6363.8 6900.1 6921.8 7003.6
0.4000904 0.0000504 0.0001769 0.3615999 0.3705429 0.6720873 0.3572373 0.2304115 0.0000955 0.0000528 0.2345543 0.0561177 0.6337368 0.2299332 0.5253794 0.0167034 0.2310706 0.208982 0.0177177 0.1437791 0.4835492 0.800003 0.0624785 0.3146795 0.0730137 0.7569417 0.2220837 0.0035902 0.0001193 0.8904281 0.6720873 0.7000186 0.0000553 0.5254785 0.1872618 0.654661 0.1596287 0.0000504 0.0295463 0.0001769 0.8087296 0.0004284 0.0844528 0.051877 0.0035668 0.2171902 0.7483488 0.1578281 0.5023196 0.0513771 0.0413297 0.0001193 0.8087296 0.0004199 0.6720873 0.2139736 0.9733515 0.7862605 0.051877 0.1101163 0.2884113 0.9012319 0.412967 0.0778071
+ + + + + + + + + + ) + + + + + + ) ) ) + + + + + + + + + ) + + + + ) + + ) + + ) ) ) + + + ) + + + ) ) + ) -
0.0002
0.0295 0.0006 0.0003 0.0117 0.0015 0.0003 0.0046 0.0040 0.0030 0.0049 0.0011 0.0091 0.0169 0.0127 0.0209
Journal of Proteome Research • Vol. 7, No. 4, 2008 1399
reviews
Callesen et al.
Table 3. Continued results in original study32
results in published study study
Traub et al.20 Traub et al.20 Traub et al.20 Traub et al.20 Traub et al.20 Traub et al.20 Villanueva et al.22 Villanueva et al.22 Villanueva et al.22 Villanueva et al.22 Villanueva et al.22 Villanueva et al.22 Villanueva et al.22 Villanueva et al.22 Villanueva et al.22 Villanueva et al.22 Villanueva et al.22 Villanueva et al.22 Villanueva et al.22 Villanueva et al.22 Villanueva et al.22 Villanueva et al.22 Villanueva et al.22 Villanueva et al.22 Vlahou et al21 Vlahou et al.21 Vlahou et al.21 Vlahou et al.21 Vlahou et al.21
m/z value [M]
m/z value [M + H]+
1060.57 1263.6 1350.64 1536.68 1739.93 1762.87 1864.95 1895.99 2115.01 2183.91 2305.2 2358.09 2379.03 2451.11 2508.16 2602.15 2659.03 2704.13 2.95 ×103 3.68 ×103 3.97 ×103 4.27 ×103 6.43 ×103
7020 8183 8605 10175 10900 11080 1061.57 1264.6 1351.64 1537.68 1740.93 1763.87 1865.95 1896.99 2116.01 2184.91 2306.2 2359.09 2380.03 2452.11 2509.16 2603.15 2660.03 2705.13
p-value
tendency
1.03 × 10-11 2.75 × 10-7 4.8 × 10-5 1.38 × 10-4 3.92 × 10-6 3.00 × 10-11 1.39 × 10-5 1.04 × 10-8 8.19 × 10-10 1.49 × 10-8 1.09 × 10-6 4.07 × 10-12 1.26 × 10-7 4.88 × 10-7 5.56 × 10-13 2.08 × 10-7 7.39 × 10-12 1.79 × 10-7
+ + + + + + + + + + + + + + + + + + +
mean m/z value [M + H]+
p-value
tendency
7018.6 8183.6 8598.3 10169.7 10896.6 11073.1 1062.3 1265.7 1352.5 1538.6 1742.6 1764.7 1867.2 1898.5 2117.3 2179.7 2307.0 2356.0 2380.3 2453.5 2507.2 2605.6 2661.6 2701.2 2953.3 3682.3 3977.4 4271.5 6430.6
0.6236014 0.8120814 0.9783533 0.5457598 0.220309 0.380048 0.176128 0.6760528 0.6137818 0.6502399 0.5986177 0.1143707 0.0048557 0.1219578 0.3831686 0.2080835 0.654661 0.800003 0.132789 0.1437795 0.5275409 0.0626867 0.99567 0.2033634 0.0000504 0.3146795 0.3705429 0.0004463 0.5614721
+ ) ) + + + + + ) + + + ) ) + ) + + ) + + + -
a Correspondence is based on (2‰ of the m/z value in the mass range up to 2.5 kDa, and 1‰ of the m/z value in the mass range from 2.5 to 25 kDa. Tendency refers to the difference in intensity of the m/z value between cases and controls (+ corresponds to higher intensity of the m/z value among cases compared to controls; - corresponds to lower intensity of the m/z value among cases compared to controls). NOTE: Only studies with a reported mass accuracy less than 100 Da are included.
We compared in our investigation only the published lists of discriminating peaks. Because of insufficient information in some of the studies, we did not include information on the degree of discriminative power of the single peaks. This may imply that we partially investigated the reproducibility of peaks, which were only of borderline significance in the original studies. In some of the studies, the sample size was small and had limited clinical details, which makes it difficult to evaluate the clinical relevance. The number and masses of discriminating peaks found also varies. In addition, the differences in cancer and control peaks could be due to inflammatory proteins, which are not necessarily cancer-specific. Moreover, as different biological matrices were used and different sample pretreatments and analysis protocols were applied in the various studies, there will be a risk that similar m/z values do not represent identical proteins. In addition, the SELDI-TOF MS mass accuracy is highly dependent on appropriate mass calibration, and even then, the accuracy can be poor. Consequently, it may be questionable whether the m/z values reported in these studies can be readily compared in the absence of specific protein identification, and the next important stages in mass spectrometry based protein profiling for diagnosis of breast cancer should be aimed at verification, validation, and identification of peaks using complementary experimental (e.g., LC-MS/MS) and clinical methods. Presently, however, these kinds of studies are not available. 1400
Journal of Proteome Research • Vol. 7, No. 4, 2008
Future Perspectives. A challenge for future studies is to provide careful reporting of preanalytical factors as sample collection, processing, and storage as it is now increasingly being recognized that preanalytical factors have a major impact on the outcome of proteomic studies.25,28–31,35–37 Standardization of sample handling is very important to obtain reproducible results, and there is a need of more comprehensive description of technical details in the future.34 It is important that data is reproducible if protein profiling is to become a diagnostic tool. Then the next important stages in mass spectrometry based protein profiling should be aimed at verification, validation, and identification of peaks using complementary methods. In relation to allowing cross-study comparisons, high quality systematic reviews, and meta-analysis, a challenge for the proteomic society could be to provide standards for reporting of this kind of studies analogue to what have been done in relation to reporting of diagnostic accuracy studies38 and in relation to reporting of randomized controlled trials.39
Conclusion We have presented the first review and comparative study of protein profiling studies of breast cancer. Despite a considerable heterogeneity between these studies and the known problems related to reproducibility, we were able to demonstrate some common features between studies. We could point
reviews
MS-Based Protein Profiles for Diagnosis of Breast Cancer to a tendency of increasing reproducibility with increasing number of studies publishing a certain peak in protein profiles. This suggests that we are converging to a set of common discriminating peaks for breast cancer, which are reproducible across different clinical studies. These potential biomarker related peaks should be further investigated and validated for future use in breast cancer research and clinical tests.
Acknowledgment. This work was supported by grants from the Danish Cancer Society, Ministry of the Interior and Health, Lands-landsdelspuljen, and a donation from Elly Jacobsen (to A.K.C.). O.N.J. is a Lundbeck Foundation Research Professor.
(16)
(17)
(18)
(19)
References (1) Cho, W. C.; Cheng, C. H. Oncoproteomics: current trends and future perspectives. Expert Rev. Proteomics 2007, 4, 401–410. (2) Diamandis, E. P. Peptidomics for cancer diagnosis: Present and future. J. Proteome Res. 2006, 5, 2079–2082. (3) Li, J.; Zhang, Z.; Rosenzweig, J.; Young, Y.; Chan, D. Proteomics and bioinformatics approaches for identification of serum biomarkers to detect breast cancer. Clin. Chem. 2002, 48, 1296–1304. (4) Becker, S.; Cazares, L. H.; Watson, P.; Lynch, H.; Semmes, O. J.; Drake, R. R.; Laronga, C. Surfaced-enhanced laser desorption/ ionization time-of-flight (SELDI-TOF) differentiation of serum protein profiles of BRCA-1 and sporadic breast cancer. Ann. Surg. Oncol. 2004, 11, 907–914. (5) Caputo, E.; Lombardi, M. L.; Luongo, V.; Moharram, R.; Tornatore, P.; Pirozzi, G.; Guardiola, J.; Martin, B. M. Peptide profiling in epithelial tumor plasma by the emerging proteomic techniques. J. Chromatogr. B: Anal. Technol. Biomed. Life Sci. 2005, 819, 59– 66. (6) Heike, Y.; Hosokawa, M.; Osumi, S.; Fujii, D.; Aogi, K.; Takigawa, N.; Ida, M.; Tajiri, H.; Eguchi, K.; Shiwa, M.; Wakatabe, R.; Arikuni, H.; Takaue, Y.; Takashima, S. Identification of serum proteins related to adverse effects induced by docetaxel infusion from protein expression profiles of serum using SELDI ProteinChip system. Anticancer Res. 2005, 25, 1197–1203. (7) Hu, Y.; Zhang, S.; Yu, J.; Liu, J.; Zheng, S. SELDI-TOF-MS: the proteomics and bioinformatics approaches in the diagnosis of breast cancer. Breast 2005, 14, 250–255. (8) Kuerer, H. M.; Coombes, K. R.; Chen, J. N.; Xiao, L.; Clarke, C.; Fritsche, H.; Krishnamurthy, S.; Marcy, S.; Hung, M. C.; Hunt, K. K. Association between ductal fluid proteomic expression profiles and the presence of lymph node metastases in women with breast cancer. Surgery 2004, 136, 1061–1069. (9) Laronga, C.; Becker, S.; Watson, P.; Gregory, B.; Cazares, L.; Lynch, H.; Perry, R. R.; Wright, G. L., Jr.; Drake, R. R.; Semmes, O. J. SELDITOF serum profiling for prognostic and diagnostic classification of breast cancers. Dis. Markers 2003–04, 19, 229–238. (10) Li, J.; Orlandi, R.; White, C. N.; Rosenzweig, J.; Zhao, J.; Seregni, E.; Morelli, D.; Yu, Y.; Meng, X. Y.; Zhang, Z.; Davidson, N. E.; Fung, E. T.; Chan, D. W. Independent validation of candidate breast cancer serum biomarkers identified by mass spectrometry. Clin. Chem. 2005, 51, 2229–2235. (11) Li, J.; Zhao, J.; Yu, X.; Lange, J.; Kuerer, H.; Krishnamurthy, S.; Schilling, E.; Khan, S. A.; Sukumar, S.; Chan, D. W. Identification of biomarkers for breast cancer in nipple aspiration and ductal lavage fluid. Clin. Cancer Res. 2005, 11, 8312–8320. (12) Mathelin, C.; Cromer, A.; Wendling, C.; Tomasetto, C.; Rio, M. C. Serum biomarkers for detection of breast cancers: A prospective study. Breast Cancer Res. Treat. 2006, 96, 83–90. (13) Mendrinos, S.; Nolen, J. D.; Styblo, T.; Carlson, G.; Pohl, J.; Lewis, M.; Ritchie, J. Cytologic findings and protein expression profiles associated with ductal carcinoma of the breast in ductal lavage specimens using surface-enhanced laser desorption and ionization-time of flight mass spectrometry. Cancer 2005, 105, 178– 183. (14) Paweletz, C. P.; Trock, B.; Pennanen, M.; Tsangaris, T.; Magnant, C.; Liotta, L. A.; Petricoin, E. F. 3rd Proteomic patterns of nipple aspirate fluids obtained by SELDI-TOF: potential for new biomarkers to aid in the diagnosis of breast cancer. Dis. Markers 2001, 17, 301–307. (15) Pawlik, T. M.; Fritsche, H.; Coombes, K. R.; Xiao, L.; Krishnamurthy, S.; Hunt, K. K.; Pusztai, L.; Chen, J. N.; Clarke, C. H.; Arun, B.; Hung, M. C.; Kuerer, H. M. Significant differences in nipple aspirate fluid protein expression between healthy women and those with breast
(20)
(21)
(22)
(23) (24)
(25)
(26)
(27)
(28)
(29) (30)
(31)
(32)
(33) (34)
cancer demonstrated by time-of-flight mass spectrometry. Breast Cancer Res. Treat. 2005, 89, 49–157. Pusztai, L.; Gregory, B. W.; Baggerly, K. A.; Peng, B.; Koomen, J.; Kuerer, H. M.; Esteva, F. J.; Symmans, W. F.; Wagner, P.; Hortobagyi, G. N.; Laronga, C.; Semmes, O. J. Pharmacoproteomic analysis of prechemotherapy and postchemotherapy plasma samples from patients receiving neoadjuvant or adjuvant chemotherapy for breast carcinoma. Cancer 2004, 100, 1814–1822. Sauter, E. R.; Zhu, W.; Fan, X. J.; Wassell, R. P.; Chervoneva, I.; Du Bois, G. C. Proteomic analysis of nipple aspirate fluid to detect biologic markers of breast cancer. Br. J. Cancer 2002, 86, 1440– 1443. Sauter, E. R.; Shan, S.; Hewett, J. E.; Speckman, P.; Du Bois, G. C. Proteomic analysis of nipple aspirate fluid using SELDI-TOF-MS. Int. J. Cancer 2005, 114, 791–796. Streckfus, C. F.; Bigler, L. R.; Zwick, M. The use of surface-enhanced laser desorption/ionization time-of-flight mass spectrometry to detect putative breast cancer markers in saliva: a feasibility study. J. Oral. Pathol. Med. 2006, 35, 292–300. Traub, F.; Feist, H.; Kreipe, H. H.; Pich, A. SELDI-MS-based expression profiling of ductal invasive and lobular invasive human breast carcinomas. Pathol. Res. Pract. 2005, 201, 763–770. Vlahou, A.; Laronga, C.; Wilson, L.; Gregory, B.; Fournier, K.; McGaughey, D.; Perry, R. R.; Wright, G. L., Jr.; Semmes, O. J. A novel approach toward development of a rapid blood test for breast cancer. Clin. Breast Cancer 2003, 4, 203–209. Villanueva, J.; Shaffer, D. R.; Philip, J.; Chaparro, C. A.; ErdjumentBromage, H.; Olshen, A. B.; Fleisher, M.; Lilja, H.; Brogi, E.; Boyd, J.; Sanchez-Carbayo, M.; Holland, E. C. Differential exoprotease activities confer tumor-specific serum peptidome patterns. J. Clin. Invest. 2006, 116, 271–284. Henderson, N. A.; Steele, R. J. SELDI-TOF proteomic analysis and cancer detection. Surgeon 2005, 3, 383–390. Baumann, S.; Ceglarek, U.; Fiedler, G. M.; Lembcke, J.; Leichtle, A.; Thiery, J. Standardized approach to proteome profiling of human serum based on magnetic bead separation and matrixassisted laser desorption/ionization time-of-flight mass spectrometry. Clin. Chem. 2005, 51, 973–980. de Noo, M. E.; Tollenaar, R. A.; Ozalp, A.; Kuppen, P. J.; Bladergroen, M. R.; Eilers, P. H.; Deelder, A. M. Reliability of human serum protein profiles generated with C8 magnetic beads assisted MALDI-TOF mass spectrometry. Anal. Chem. 2005, 77, 7232–7241. Villanueva, J.; Philip, J.; Entenberg, D.; Chaparro, C. A.; Tanwar, M. K.; Holland, E. C.; Tempst, P. Serum peptide profiling by magnetic particle-assisted, automated sample processing and MALDI-TOF mass spectrometry. Anal. Chem. 2004, 76, 1560–1570. Callesen, A. K.; Mohammed, S.; Bunkenborg, J.; Kruse, T. A.; Cold, S.; Mogensen, O.; Christensen, R.; Vach, W.; Jørgensen, P. E.; Jensen, O. N. Serum protein profiling by miniaturized solid-phase extraction and matrix-assisted laser desorption/ionization mass spectrometry. Rapid Commun. Mass Spectrom. 2005, 19, 1578–1586. Baggerly, K. A.; Morris, J. S.; Coombes, K. R. Reproducibility of SELDI TOF protein patterns in serum: comparing data sets from different experiments. Bioinformatics 2004, 20, 777–785. Boguski, M. S.; McIntosh, M. W. Biomedical informatics for proteomics. Nature 2003, 422, 233–237. Somorjai, R. L.; Dolenko, B.; Baumgartner, R. Class prediction and discovery using gene microarray and proteomics mass spectroscopy data: curses, caveats, cautions. Bioinformatics 2003, 19, 1484– 1491. Yasui, Y.; Pepe, M.; Thompson, M. L.; Adam, B. L.; Wright, G. L., Jr.; Qu, Y.; Potter, J. D.; Winget, M.; Thornquist, M.; Feng, Z. A data-analytic strategy for protein biomarker discovery: profiling of high-dimensional proteomic data for cancer detection. Biostatistics 2003, 4, 449–463. Callesen, A. K.; Vach, W.; Jørgensen P. E.; Cold, S.; Tan, Q.; Christensen, R.; Mogensen, O.; Kruse, T. A.; Jensen, O. N.; Madsen, J. S. Combined experimental and statistical strategy for mass spectrometry based serum protein profiling for diagnosis of breast cancer: A case-control study. J. Proteome Res. 2008, 7, 1419–1426. Coombes, K. R. Analysis of mass spectrometry profiles of the serum proteome. Clin. Chem. 2005, 51, 1–2. Bossuyt, P. M.; Reitsma, J. B.; Bruns, D. E.; Gatsonis, C. A.; Glasziou, P. P.; Irwig, L. M.; Lijmer, J. G.; Moher, D.; Rennie, D.; de Wet, H. C. W. Towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative. Br. Med. J. 2003, 326, 41–44.
Journal of Proteome Research • Vol. 7, No. 4, 2008 1401
reviews (35) Bons, J. A. P.; Wodzig, W. K. W. H.; van Dieijen-Visser, M. P. Protein profiling as a diagnostic tool in clinical chemistry: a review. Clin. Chem. Lab. Med. 2005, 43, 1281–1290. (36) Rai, J. A.; Gelfand, C. A.; Haywood, B. C.; Warunek, D. J.; Yi, J.; Schuchard, M. D.; Mehigh, R. J.; Cockrill, S. L.; Scott, G. B. I.; Tammen, H.; Sculz-Knappe, P.; Speicher, D. W. HUPO Plasma Proteome Project specimen collection and handling: Towards the standardization of parameters for plasma proteome samples. Proteomics 2005, 5, 3262–3277. (37) Sorace, J. M.; Zhan, M. A data review and re-assessment of ovarian cancer serum proteomic profiling. BMC Bioinf. 2003, 4, 24.
1402
Journal of Proteome Research • Vol. 7, No. 4, 2008
Callesen et al. (38) Bossuyt, P. M.; Reitsma, J. B.; Bruns, D. E.; Gatsonis, C. A.; Glasziou, P. P.; Irwig, L. M.; Lijmer, J. G.; Moher, D.; Rennie, D.; de Wet, H. C. W. Towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative. Br. Med. J. 2003, 326, 41–44. (39) Moher, D.; Jones, A.; Lepage, L. Use of the CONSORT statement and quality of reports of randomized trials: a comparative before-and-after evaluation. JAMA, J. Am. Med. Assoc. 2001, 285, 1992–1995.
PR800115F
