Recent years have witnessed the growing use of mass spectrometry (MS) in the clinical laboratory. MS provides massive improvements in the sensitivity and specificity of clinical tests. It does this by using an ionized molecule’s mass/charge (m/z) ratio for identification.
MS has its roots in the screening and fingerprinting of molecules in drugs of abuse. Over the years, the technology has rapidly evolved. Today, it is routinely used to screen for diseases and to precisely identify causes of infections for targeted therapies. The analysis of proteins is also accelerating, with special potential demonstrated by biomarkers such as thyroglobulin.Alongside, limitations to immunoassays have also driven adoption of MS. For example, no immunoassays were approved for the immunosuppressant sirolimus and this compelled laboratories to turn to MS. Another advantage lay in improved assay quality, such as the measurement of testosterone in patients with low endogenous concentrations, such as women and children.

Enticing advantages
One of the most enticing set of advantages of mass spectrometry is that it provides clinically relevant information from relatively small sample volumes, and does this both rapidly and at a reduced cost. Gas chromatography (GC), liquid chromatography (LC) and ion mobility spectrometry (IMS) separation now allow targeting of ever-smaller analyte concentrations. LC-MS/MS (liquid chromatography-tandem mass spectrometry), on its part, offers scope to cut costs further, while continuing to improve accuracy. Other technology trends include integrating MS with low-flow chromatography, ultra-high pressure chromatography and online/multi-dimensional  chromatography.
Nevertheless, MS also adds a new layer of complexity. As a result, close and well-structured communication between laboratories and clinicians is a vital component for the effective use of MS.

A three-step process
Today, there are three principal steps for conducting an analysis by MS: sample preparation, separation by gas-chromatography (GC) or liquid-chromatography (LC), and mass spectrometric analysis.
In MS, a sample ‘matrix’ refers to everything present in a sample, excluding analytes of interest. Differences in behaviour between analytes and matrix components determines the choice of sample preparation. Although sample preparation requires more labour than immunoassays, in-house mass spectrometry-based assays are now considered cost-effective, even for smaller labs.

Sample preparation
The preparation of samples and their subsequent separation by chromatography both use mechanisms which first position molecules (the stationary phase) and then separate analytes from matrix components (the mobile phase).
Preparation firstly depends on the sample type selected for analysis (e.g. blood/serum or urine). Analytes from serum (including blood fractions) require the maximum care in preparation, owing to a relatively low ratio in the concentration of analytes to matrix components. On the other hand, urine analytes are often compatible with simple dilution protocols, due to the concentrating effect of kidneys in the production of urine.

Typical techniques in preparing samples include solid-phase extraction (SPE), immunoextraction and dilution. The choice depends principally on whether the analytes are acidic or alkaline, and if they are heavily protein-bound.

Solid-phase extraction
SPE is based on combining a solid stationary phase with a liquid mobile phase.
Analytes of interest (and matrix components) remain in the liquid phase or associate only temporarily with the solid stationary phase. The amount of time taken up by the latter is based on characteristics such as charge and polarity of the analytes versus matrix components. A binding-and-wash solvent (different from the elution solvent), provides a relatively crude separation of analytes from the (unwanted) components.

Immunoextraction
Immunoextraction (also known as immunoaffinity purification) is based on the use of antibodies in a solid phase. This separates antibody-bound analytes from free matrix components.

Dilution
When analytes are present in high concentrations, dilution provides a simple and effective methodology to reduce matrix components. Dilution (often called ‘dilute-and-shoot’) is a common method of sample preparation for comprehensive screening and for confirmatory testing for drugs in urine.

Separation
Gas chromatography
GC chromatography uses hydrogen or helium to push molecules into a column (known as the stationary phase). Modifying the column temperature then changes the affinity of molecules in the stationary phase, thereby separating analytes from matrix components (known as the mobile phase). Though largely relevant for volatile, heat-stable compounds, ‘derivatization’ via chemical modification can increase compatibility with GC.
GC mass spectrometry (GC-MS) remains the most common method for comprehensive drug screening in the clinical laboratory.

Liquid chromatography
LC chromatography is largely used for separation of samples before MS analysis. This is largely due to a wide range of LC-compatible analytes and a reduced need for derivatization. The mobile phase in LC uses a combination of organic solvents and water. Adjustments to the ratio between the organics and water redistributes components between the mobile and stationary phases.

Ionization techniques: APCI and electrospray
MS detects charged analytes in the gaseous phase alone. Ionization is required to convert liquid-phase analytes for analysis. The two most common methods in the clinical laboratory consist of atmospheric pressure chemical ionization (APCI) and electrospray ionization.
APCI produces ions by using heat to evaporate the solvent and plasma to ionize the sample. Physical interaction with gaseous analytes leads to formation of negative or positive ions.
Electrospray ionization, on its part, combines electricity, air and heat to produce successively smaller and concentrated droplets from the liquid which elutes off a chromatographic column. This leads to a dramatic increase in charge per unit volume. Ions on the droplet surface desorb from liquid to gas phase, and the latter is introduced into the mass spectrometer.

Sample transfer to mass spectrometer
There are several choices for introducing samples into a mass spectrometer. These range from direct infusion to multidimensional chromatographic separation. The latter enables the staggered delivery of analytes and matrix components. This permits more effective utilization of analyser time by limiting analysis to fractions containing analytes of interest.

Methods of analysis
MS analysis is largely based on quadrupole analysers, time-of-flight (TOF) analysers and tandem mass spectrometers, as well as combinations of the three.

Quadrupole analysers
Linear quadrupole analysers are currently the most common type of mass spectrometer in a clinical laboratory. Called quadrupoles due to the presence of four parallel rods in a square, one pair of (diagonally opposed) rods is positively charged, while the other is negatively charged. The charges are optimized and alternated based on the analyte of specific interest. Via sequential attraction and repulsion, an ion of interest can be programmed to maintain a stable flight path between the rods. The charge and frequency can moreover be rapidly altered to sequentially detect different analytes. Quadrupole analysers have high sensitivity and mass accuracy. On the other hand, they have a limited range in mass/charge (m/z) ratios – which, as noted previously, is a unique identifier for a particular ion.

Time-of-flight analysers
Time-of-flight (TOF) mass spectrometers are based on using an electric field which accelerates gas phase ions to a detector. The time taken for this travel is based on an ion’s m/z ratio, with low m/z ions travelling faster than higher ones.
TOF analysers have an essentially unlimited m/z range and high sensitivity and accuracy, but users face limits in their dynamic range.

Key challenges in clinical MS
Tandem mass spectrometry
Successful identification of an analyte by m/z alone does not always confer specificity. One good example is morphine and hydromorphone. Though the two are distinct, they have identical positive ions, with 286 m/z.
Tandem mass spectrometers (MS/MS) use multiple quadrupoles in series. One typical configuration is to use three quadrupoles. The first and third (denoted Q1 and Q3) use combinations of charge and frequencies as described above (see section on ‘Quadrupole Analysers’). The second quadrupole, denoted q2 (in smaller case), serves as a collision cell with an inert gas (e.g. nitrogen). On entry into q2, ions collide with the inert gas, and fragment into smaller product ions which then pass through Q3 and hit a detector.
In the case of morphine and hydromorphone, q2 entry produces stable product ions (m/z of 153 for the former and 157 for the latter). After this, setting the Q3 charge and frequency to first transmit the product ion for morphine and then change the charge/frequency settings to transmit the  hydromorphone product ion results in a way to measure and differentiate between the two compounds. This approach is also known as multiple reaction monitoring (MRM) and allows a mass spectrometer to scan faster – by targeting specific m/z set points rather than a broad range.

Ion suppression/enhancement and ion ratios
Ion suppression and enhancement are two of the most common problems facing MS. These occur when a substance in a sample interferes with the ionization process of the analytes. These can range from matrix molecules to co-eluting compounds. For example, components in the sample with lower volatility can reduce the efficiency of solvent evaporation, resulting in reduced ion formation.
There are several options to reduce or eliminate such interference, including mobile phase additives to aid ionization.
Another approach is to use ion ratios. When analytes of interest are present alongside structurally similar compounds in complex matrices, interference risks rise due to co-eluting molecules with identical mass. However, ion ratios seek to monitor multiple m/z transitions for each analyte and determine ratios of chromatographic peak area for more abundant fragments to less abundant ones. The use of ion ratios further enhances the specificity of MS.

The future
Endocrinology
Although industry has sought to use antibody-mediated detection to overcome the inherent limitations of immunoassays in identifying proteins and small molecules, these have yet to be meaningfully eradicated.
Rising interest in (regular and more-frequent) testing for vitamin D has also driven implementation of LC-MS/MS (liquid chromatography-tandem mass spectrometry), which separates vitamin D2 from vitamin D3 and provide information on its epimeric form. This is not possible with existing immunoassays.
Meanwhile, although steroid hormone assays for diagnostic and forensic testing continue to grow, a lack of specificity and accuracy at low concentrations has hampered the diagnosis of endocrine disorders. This has led several medical groups to recommend mass spectrometry as the preferred method of analysis, in spite of the high degree of technical competence, skill and experience required to achieve meaningful results.

Metabolomics
Measurements of the genome and proteome need to be accompanied by quantified data on the metabolome to comprehend differences between disease and healthy status, and provide meaningful diagnosis and monitoring of disease. One of the fastest growing areas for MS in metabolomics is the screening of newborns.

Protein analysis
The success of MS in precise measurement of small molecules has driven interest in using it for peptide and protein analysis for diagnostic testing. In spite of some challenges, quantitative proteomics (covering factors such as isotope dilution and m/z transitions) is an especially exciting application for mass spectrometry.

Alterations of the microbiome are associated with colorectal cancer. Research suggests that microbiome data could improve colorectal cancer screening. Analysis of the microbiome directly from existing screening methods offers the opportunity to rapidly translate this research into practice, with the potential to develop a multifactorial colorectal cancer screening tool.

by Dr Caroline Young and Professor Philip Quirke

Current colorectal cancer screening methods
Different countries have adopted various approaches to colorectal cancer screening. They share a common goal: detection of asymptomatic adenomas or early stage carcinomas, as detection and treatment at an earlier stage is associated with improved survival [1]. Two main screening methods are in use: detection of fecal occult blood and visualization of the colon. Stool DNA testing has recently been approved but is currently prohibitively expensive.

Detection of fecal occult blood can be achieved using the guaiac fecal occult blood test (gFOBT) or an immunochemical method, fecal immunochemical test (FIT). The gFOBT method requires participants to apply stool to a gFOBT card on three occasions and return this to a screening centre through the post. Hydrogen peroxide is applied and if heme is present, blue discolouration occurs. This method has been shown to reduce mortality by 16 % [2]. The FIT method requires participants to insert a FIT probe into stool and return this to a screening centre through the post. An antibody-based assay is used to detect globin. FIT is more sensitive and specific, can be analysed quantitatively and has improved acceptability [3]. Participants in whom fecal occult blood is detected above a threshold, by either method, are referred for colonoscopy.

Alternatively, direct visualization of the colon by colonoscopy/sigmoidoscopy can be undertaken as first-line screening. Limitations include procedural risks, associated costs, workforce capacity and reduced acceptability [4].

The microbiome and colorectal cancer
The microbiome can be characterized using a number of technologies: next generation sequencing (NGS) of bacterial 16SrRNA, whole genome shotgun metagenomics of bacterial communities or the analysis of fecal metabolites (metabolomics). These techniques have enabled an appreciation of the diversity and function of the microbiome in health and disease.

Epidemiological studies demonstrate that the incidence of colorectal cancer is highest in countries with a Western culture, which encompasses Western diet, sanitation and hygiene, medication use, urbanization, etc. [5]. Migrant populations to such countries acquire the increased risk, suggesting an environmental risk factor. African Americans, who typically have a high incidence of colorectal cancer, have been shown to have different microbiomes to Native Africans, who have a low incidence of colorectal cancer [6] and the diets typical of these two groups have been shown to differentially influence the microbiome [7].

Numerous studies have found differences in the microbiome, ‘dysbiosis’, of patients with colorectal adenomas or carcinomas compared to healthy controls [8]. In general, dysbiosis is characterized by a decrease of short chain fatty acid-producing bacteria, an increase of bacteria that produce bile salts or hydrogen sulphide, an increase of pathogenic bacteria and inflammation [9]. In particular, the species Fusobacterium nucleatum, a Gram-negative oral commensal, has been associated with colorectal carcinoma in many studies.

Animal models have explored potential mechanisms [10] and interestingly show that risk is transferable with transplant of dysbiotic microbiomes. This suggests that dysbiosis may be causative or promotional of the development of colorectal cancer, rather than merely associative.

Given the association between dysbiosis and colorectal cancer, researchers have considered whether the microbiome could be used as a screening tool.

The microbiome compared to gFOBT
Several studies have compared the accuracy of the microbiome as a screening tool to gFOBT. Amiot et al. showed that a screening model combining age plus microbiome (typed by qPCR) was no better than a model combining age plus gFOBT [11]. However, metabolomic analysis [by 1(H)-NMR spectroscopy] was more accurate than gFOBT [12]. Zeller et al. created a screening model that combined metagenomic data with gFOBT results, which lead to an increase in sensitivity compared to gFOBT alone. This model was subsequently validated in a cohort of a different nationality. It showed some ability to distinguish colorectal cancer from a distinct bowel condition (inflammatory bowel disease) and could be extrapolated to NGS of 16SrRNA (a cheaper method) [13].

Zackular et al. used 16SrRNA analysis of the microbiome to create models combining microbiome data and patient metadata that were more accurate than models based on metadata alone [14]. A model comprising BMI, microbiome data and gFOBT was more accurate at distinguishing adenoma from carcinoma than gFOBT alone. Yu et al. used metagenomics to identify two discriminatory bacterial genes that they then validated as biomarkers by qPCR (a cheaper method) in a cohort of a different nationality. The area under the receiver operating characteristic (ROC) curve for discriminating carcinoma from controls was 0.84, although gFOBT or FIT screening was not performed for comparison [15].

The microbiome compared to FIT
As FIT is replacing gFOBT in many screening programmes and has a higher sensitivity, comparing the accuracy of the microbiome as a screening tool with FIT is more appropriate.

Baxter et al. used 16SrRNA to create a screening model that combined microbiome data and FIT to discriminate healthy controls from cases with either adenoma or carcinoma [16]. This model was more sensitive but less specific than FIT alone; it detected 70% of cancers and 37% of adenomas which were missed by FIT. Liang et al. [17] identified four bacterial species (one being F. nucleatum) by qPCR that could distinguish colorectal carcinoma from healthy controls with greater accuracy than FIT. Combining microbiome and FIT data afforded greater accuracy still.

Goedert et al. [18] analysed the microbiome by 16SrRNA in patients with a positive FIT result at baseline. The microbiome data gave an area under the ROC curve for discriminating between healthy controls and colorectal adenoma of 0.767.

Limitations of current research
The studies mentioned above show promise for the microbiome as a potential colorectal cancer screening tool. However, they should be interpreted with a degree of caution, owing to a number of limitations which mean that aspects of the studies do not realistically reflect screening conditions. Several of the studies assessed participants at increased risk of colorectal cancer or who were symptomatic. Some collected stool samples following bowel preparation and colonoscopy; one study found that this did not affect the significance of results [16], whereas another found that it did [15]. Several studies included adenomas <10 mm within their control groups. Many of the studies created models that distinguished adenomas from carcinomas or carcinomas from healthy controls; few designed models to discriminate between healthy controls and participants with any colorectal lesion (i.e. either adenoma or carcinoma).

All of the studies used whole stool samples that were refrigerated or frozen by participants at home or delivered within a limited time window to research centres. This method of sample collection would not translate to national screening programmes, which already struggle with poor participant uptake. In light of this, researchers have, therefore, investigated whether the microbiome can be analysed directly from the existing screening tools, gFOBT or FIT.

Analysing the microbiome directly from existing screening tools
Sinha et al. emphasize the need to assess reproducibility, stability over time and how accurately results reflect the gold standard (fresh or immediately frozen stool) when analysing different methods of microbiome sample collection [19]. They found that 16SrRNA microbiome results were similar when analysed from unprocessed or processed gFOBT cards and, in addition to Dominianni et al. [20], showed stability after storage at room temperature for several days. This work was extended by Taylor et al. [21] who demonstrated that the microbiome is stable when analysed by 16SrRNA from processed gFOBT cards stored at room temperature for up to 3 years.

Lotfield et al. showed that metabolomic assessment of the microbiome by ultra-performance liquid chromatography and high resolution/tandem mass spectrometry was stable and accurate (albeit with a degree of bias affecting certain metabolite groups) when analysed directly from gFOBT samples but not from FIT samples [22]. This suggests that different methods of sample collection may be more or less appropriate dependent upon the method of microbiome analysis.

These studies have assessed methods of microbiome sample collection from healthy volunteers. Baxter et al. [23] have analysed the microbiome directly from processed FIT from subjects with normal bowels, colorectal adenomas or carcinomas. Their study comes with the caveat that some of the stool samples were collected after bowel preparation and colonoscopy; samples were stored at −80 °C before being thawed and transferred to FIT; FIT was refrigerated for up to 2 days, processed, then stored at −20 °C before being thawed for microbiome analysis. The study demonstrated that a screening model to discriminate between healthy controls and subjects with any colonic lesion had a similar area under the ROC curve whether microbiome analysis was performed directly from FIT samples or whole stool samples.

As an alternative to stool, Westenbrink et al. analysed microbiome-related volatile organic compounds from urine [24] and described a similar sensitivity for the detection of colorectal cancer as gFOBT or FIT.

Conclusion
Research suggests that there is potential for microbiome analysis to both augment and to be integrated with existing screening methods. The landscape of colorectal cancer screening is changing [25]; it seems likely that a more sophisticated, multifactorial screening tool will be adopted. Microbiome analysis is likely to contribute and may even offer information beyond that of screening, e.g. prevention or treatment targets [26]. Furthermore, collection of longitudinal, population-based microbiome data via national screening programmes will transform the field of microbiome research.

References
1. Cancer Research UK (http://www.cancerresearchuk.org).
2. Hewitson P, Glasziou PP, Irwig L, Towler B, Watson E. Screening for colorectal cancer using the faecal occult blood test, Hemoccult. Cochrane Database Syst Rev. 2007; DOI: 10.1002/14651858.CD001216.pub2
3. Schreuders EH, Grobbee EJ, Spaander MC, Kuipers EJ. Advances in fecal tests for colorectal cancer screening. Curr Treat Options Gastroenterol. 2016; 14(1): 152–162.
4. US Preventive Services Task Force, Bibbins-Domingo K, Grossman DC, Curry SJ, Davidson KW, Epling JW Jr, García FA, Gillman MW, Harper DM, et al. Screening for colorectal cancer: US preventive services task force recommendation statement. JAMA 2016; 315(23): 2564–2575.
5. Haggar FA, Boushey RP. colorectal cancer epidemiology: incidence, mortality, survival, and risk factors. Clin Colon Rectal Surg. 2009; 22(4): 191–197.
6. Ou J, Carbonero F, Zoetendal EG, DeLany JP, Wang M, Newton K, Gaskins HR, O’Keefe SJ. Diet, microbiota, and microbial metabolites in colon cancer risk in rural Africans and African Americans. Am J Clin Nutr. 2013; 98(1): 111–120.
7. David LA, Maurice CF, Carmody RN, Gootenberg DB, Button JE, Wolfe BE, Ling AV, Devlin AS, Varma Y, et al. Diet rapidly and reproducibly alters the human gut microbiome. Nature 2014; 505(7484): 559–563.
8. Borges-Canha M, Portela-Cidade JP, Dinis-Ribeiro M, Leite-Moreira AF, Pimentel- Nunes P. Role of colonic microbiota in colorectal carcinogenesis: a systematic review. Rev Esp Enferm Dig. 2015; 107(11): 659–671.
9. Sun J, Kato I. Gut microbiota, inflammation and colorectal cancer. Genes Dis. 2016; 3(2): 130–143.
10. Keku TO, Dulal S, Deveaux A, Jovov B, Han X. The gastrointestinal microbiota and colorectal cancer. Am J Physiol Gastrointest Liver Physiol. 2015; 308(5): G351–363.
11. Amiot A, Mansour H, Baumgaertner I, Delchier JC, Tournigand C, Furet JP, Carrau JP, Canoui-Poitrine F, Sobhani I; CRC group of Val De Marne. The detection of the methylated Wif-1 gene is more accurate than a fecal occult blood test for colorectal cancer screening. PLoS One 2014; 9(7): e99233.
12. Amiot A, Dona AC, Wijeyesekera A, Tournigand C, Baumgaertner I, Lebaleur Y, Sobhani I, Holmes E. (1)H NMR spectroscopy of fecal extracts enables detection of advanced colorectal neoplasia. J Prot Res. 2015; 14(9): 3871–3881.
13. Zeller G, Tap J, Voigt AY, Sunagawa S, Kultima JR, Costea PI, Amiot A, Böhm J, Brunetti F, et al. Potential of fecal microbiota for early-stage detection of colorectal cancer. Mol Syst Biol. 2014; 10: 766.
14. Zackular JP, Rogers MA, Ruffin MT 4th, Schloss PD. The human gut microbiome as a screening tool for colorectal cancer. Cancer Prev Res (Phila). 2014; 7(11): 1112–1121.
15. Yu J, Feng Q, Wong SH, Zhang D, yi Liang Q, Qin Y, et al. Metagenomic analysis of faecal microbiome as a tool towards targeted non-invasive biomarkers for colorectal cancer. Gut 2015; DOI: 10.1136/gutjnl-2015-309800.
16. Baxter NT, Ruffin MT 4th, Rogers MA, Schloss PD. Microbiota-based model improves the sensitivity of fecal immunochemical test for detecting colonic lesions. Genome Med. 2016; 8(1): 37.
17. Liang JQ, Chiu J, Chen Y, Huang Y, Higashimori A, Fang JY, Brim H, Ashktorab H, Ng SC, et al. Fecal bacteria act as novel biomarkers for non-invasive diagnosis of colorectal cancer. Clin Cancer Res. 2016; DOI: 10.1158/1078-0432.CCR-16-1599.
18. Goedert JJ, Gong Y, Hua X, Zhong H, He Y, Peng P, Yu G, Wang W, Ravel J, et al. Fecal microbiota characteristics of patients with colorectal adenoma detected by screening: a population-based study. EBioMedicine 2015; 2(6): 597–603.
19. Sinha R, Chen J, Amir A, Vogtmann E, Shi J, Inman KS, Flores R, Sampson J, Knight R, Chia N. Collecting fecal samples for microbiome analyses in epidemiology studies. Cancer Epidemiol Biomarkers Prev. 2016; 25(2): 407–416.
20. Dominianni C, Wu J, Hayes RB, Ahn J. Comparison of methods for fecal microbiome biospecimen collection. BMC Microbiol. 2014; 14: 103.
21. Taylor M, Wood H, Halloran S, Quirke P. Examining the potential use and long term stability of guaiac faecal occult blood test cards for microbial DNA 16srRNA sequencing. J Clin Pathol. Accepted for publication.
22. Loftfield E, Vogtmann E, Sampson JN, Moore SC, Nelson H, Knight R, Chia N, Sinha R. Comparison of collection methods for fecal samples for discovery metabolomics in epidemiologic studies. Cancer Epidemiol Biomarkers Prev. 2016; 25(11): 1483–1490.
23. Baxter NT, Koumpouras CC, Rogers MA, Ruffin MT 4th, Schloss P. DNA from fecal immunochemical test can replace stool for microbiota-based colorectal cancer screening. Microbiome 2016; 4(1): 59.
24. Westenbrink E, Arasaradnam RP, O’Connell N, Bailey C, Nwokolo C, Bardhan KD, Covington JA. Development and application of a new electronic nose instrument for the detection of colorectal cancer. Biosens Bioelectron. 2015; 67: 733–738.
25. Nguyen MT, Weinberg DS. Biomarkers in colorectal cancer screening. J Natl Compr Canc Netw. 2016; 14(8): 1033–1040.
26. Pitt JM, Vetizou M, Waldschmitt N, Kraemer G, Chamaillard M, Boneca IG, Zitvogel L. Fine-tuning cancer immunotherapy: optimizing the gut microbiome. Cancer Research 2016; 76(16): 4602–4607.

The authors
Caroline Young* MA, BMBCh; Philip Quirke BM, PhD, FRCPath, FMedSci
Wellcome Trust Brenner Building, St James University Hospital, Leeds LS9 7TF, UK

*Corresponding author
E-mail: caroline.young4@nhs.net

Ovarian cancer is difficult to diagnose early, with consequent poor survival. Evidence suggests many cases may originate in precursor lesions in the fallopian tubes. Differential expression of specific proteins in the fallopian tubes of women with high-grade serous ovarian cancer, detected by immunohistochemistry, shows promise as a potential novel diagnostic marker.

by Dr Kezia Gaitskell and Prof. Ahmed Ashour Ahmed

Background
Ovarian cancer is the seventh greatest cause of cancer mortality amongst women worldwide, and the fifth greatest cause amongst women in more developed regions [1]. In the USA, 60% of women with ovarian cancer already have distant metastases at diagnosis, for which the 5-year survival is less than 30% [2]. Early clinical diagnosis of ovarian cancer is difficult, as symptoms are often non-specific, such as abdominal distention, urinary frequency, or abdominal pain [3].

Current evidence on ovarian cancer diagnosis and screening
Diagnostic investigations include imaging (e.g. ultrasound, CT or MRI of the pelvis and abdomen), together with blood tests for tumour markers – particularly cancer antigen 125 (CA-125) [4]. However, although CA-125 is the main biomarker used in the diagnosis of ovarian cancer, it is far from perfect in sensitivity and specificity: although approximately 80% of women with epithelial ovarian cancer will have a CA-125 concentration above the cut-off value of 35 IU/mL, CA-125 may also be elevated with other cancers (including liver, pancreatic, lung, and endometrial cancers), and physiological or benign conditions (including menses, pregnancy, cirrhosis, salpingitis, pancreatitis and endometriosis) [5].

A variety of additional putative tumour markers have been suggested for use in combination with CA-125, but current guidelines from the National Comprehensive Cancer Network in the USA are that there is insufficient evidence for their usefulness in detecting early-stage ovarian cancer [4], although the European Group on Tumor Markers suggests that human epididymis protein 4 (HE4) may be helpful for the differential diagnosis of pelvic masses, particularly in premenopausal women [5].

There has been considerable interest in finding markers of early disease, that could enable earlier diagnosis or screening for ovarian cancer. Two large randomized controlled trials have been performed of screening for ovarian cancer: the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial (PLCO) in the USA [6], and the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS) in the UK [7]. Both trials randomized women to either no screening, or screening with the CA-125 blood test with or without trans-vaginal ultrasound. Unfortunately, neither trial could demonstrate a clear mortality benefit with screening, although there was a suggestion of benefit in some secondary analyses in the UKCTOCS trial [7].

New hypotheses of the origins of ovarian cancer
The search for potential early markers of ovarian cancer is also affected by the increasing evidence of heterogeneity between the tumour subtypes. Ovarian cancer has traditionally been divided into subtypes on the basis of microscopic morphology, the most common types being serous, endometrioid, clear cell, and mucinous tumours. There is growing evidence that these different histological tumour subtypes have different characteristic genetic mutations, and may have distinct origins [8, 9]. In particular, there is evidence that many cases of high-grade serous ovarian cancer (the most common subtype) may arise from precursor lesions in the fallopian tube epithelium, known as serous tubal intraepithelial carcinoma (STIC) (reviewed by Nik et al. [10]). These STIC lesions show dysplastic morphological changes, and also tend to show the mutations in the tumour-suppressor gene TP53 that are characteristic of high-grade serous ovarian carcinoma, and increased expression of the proliferation marker Ki-67. There is also evidence that some cases of endometrioid and clear cell ovarian cancer (less common subtypes) may arise from endometriosis (reviewed by Munksgaard & Blaakaer [11]). The origins of low-grade serous tumours and mucinous carcinomas are less certain, although several hypotheses exist.

The hypothesis that many, if not most, high-grade serous ‘ovarian’ cancers may in fact arise from the fallopian tubes has led to increasing interest in exploring changes in the fallopian tubes as potential early markers. The discovery of STIC lesions is interesting in terms of improving our understanding of pathogenesis, but is not currently useful for identifying changes early in malignancy, or pre-malignancy, in clinical practice. One limitation is that STIC lesions tend to be very focal, and are most common at the fimbrial end of the fallopian tubes, adjacent to the ovary, which is difficult to access without surgical removal of the fallopian tubes.

New findings regarding the role of SOX2
We investigated increased expression of SOX2, a key stem cell differentiation gene, as a possible marker of high-grade serous carcinogenesis within the fallopian tubes. We chose SOX2 because work from our group had shown that mutations at several sites near the SOX2 gene were ubiquitous in samples of ovarian cancer taken from multiple locations and time points in a single patient, indicating that they acted as early ‘driver’ mutations [12]. We showed that SOX2 expression (detected using immunohistochemistry) was significantly increased in the fallopian tube epithelial cells of women with high-grade serous ovarian cancer, compared to women with endometrial cancer or benign disease (e.g. uterine fibroids) [12], as illustrated in Figure 1.

We also found that SOX2 expression in the fallopian tubes was significantly increased in women with germline mutations in the tumour suppressor genes BRCA1 and BRCA2, who are known to be at higher risk of breast and ovarian cancer [12]. These women with BRCA1/2 mutations had their ovaries and fallopian tubes removed to reduce their subsequent risk of cancer, but did not have evident cancer at the time of surgery. Thus, the finding that elevated SOX2 expression was detectable in their fallopian tubes suggests that increased SOX2 expression may be an early sign of precancerous changes within the fallopian tubes.

Potential future implications
Our observation that SOX2 expression is increased in the fallopian tube epithelial cells of women with high-grade serous ovarian cancer, and women with BRCA1/2 mutations, compared to women with other cancers or benign disease, suggests that SOX2 might have a potential role as a biomarker in the early diagnosis of ovarian cancer. However, several challenges remain before testing for SOX2 expression could be considered in clinical practice – particularly the anatomical difficulty of sampling the fallopian tube epithelium without invasive surgery, and the fact that SOX2 is a nuclear marker. Our research group is currently exploring other potential cell-surface markers that correlate with SOX2 expression, which might be easier to detect.

Summary
There are many challenges in the early diagnosis of ovarian cancer. New evidence of the possible tubal origins of high-grade serous ovarian cancer is changing the approach to identifying potential new biomarkers of early disease. SOX2 has emerged as a promising marker, but further work is needed before it would be suitable for routine clinical practice.

References
1. Ferlay J, Soerjomataram I, et al. GLOBOCAN 2012 v1.0: Estimated cancer incidence and mortality worldwide in 2012. International Agency for Research on Cancer/World Health Organization 2013 (http://globocan.iarc.fr/Default.aspx).
2. Howlader N, Noone AM, et al. Cronin KA (eds). SEER Cancer Statistics Review, 1975-2013. National Cancer Institute, Bethesda, MD, USA 2016 (http://seer.cancer.gov/csr/1975_2013/).
3. Hamilton W, Peters TJ, et al. Risk of ovarian cancer in women with symptoms in primary care: population based case-control study. BMJ 2009; 339: b2998.
4. National Comprehensive Cancer Network. NCCN Clinical Practice Guidelines in Oncology (NCCN Guidelines): Ovarian cancer including fallopian tube cancer and primary peritoneal cancer, Version 1.2016. Ft. Washington, PA, USA. National Comprehensive Cancer Network 2016 (https://www.nccn.org/professionals/physician_gls/pdf/ovarian.pdf).
5. Soletormos G, Duffy MJ, et al. Clinical use of cancer biomarkers in epithelial ovarian cancer: updated guidelines from the European group on tumor markers. Int J Gynecol Cancer 2016; 26(1): 43–51.
6. Buys SS, Partridge E, et al. Effect of screening on ovarian cancer mortality: the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Randomized Controlled Trial. JAMA 2011; 305(22): 2295–2303.
7. Jacobs IJ, Menon U, et al. Ovarian cancer screening and mortality in the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS): a randomised controlled trial. Lancet 2016; 387(10022): 945–956.
8. Kurman RJ, Shih IeM. Molecular pathogenesis and extraovarian origin of epithelial ovarian cancer–shifting the paradigm. Hum Pathol. 2011; 42(7): 918–931.
9. Prat J. Ovarian carcinomas: five distinct diseases with different origins, genetic alterations, and clinicopathological features. Virchows Arch. 2012; 460(3): 237–249.
10. Nik NN, Vang R, et al. Origin and pathogenesis of pelvic (ovarian, tubal, and primary peritoneal) serous carcinoma. Annu Rev Pathol. 2014; 9: 27–45.
11. Munksgaard PS, Blaakaer J. The association between endometriosis and ovarian cancer: a review of histological, genetic and molecular alterations. Gynecol Oncol. 2012; 124(1): 164–169.
12. Hellner K, Miranda F, et al. Premalignant SOX2 overexpression in the fallopian tubes of ovarian cancer patients: discovery and validation studies. EBioMedicine 2016; 10: 137–149.

The authors
Kezia Gaitskell*^1,2 BM BCh; Ahmed Ashour Ahmed^1,2 MBBCh, MRCOG, PhD
¹Ovarian Cancer Cell Laboratory, Weatherall Institute of Molecular Medicine, University of Oxford, Headington, Oxford OX3 9DS, UK
²Nuffield Department of Obstetrics and Gynaecology, University of Oxford, Women’s Centre, John Radcliffe Hospital, Oxford OX3 9DU, UK

*Corresponding author
E-mail: Kezia.gaitskell@ceu.ox.ac.uk