I have always wanted to quote Miranda (a very famous UK comedy show aired on BBC). In that show, Miranda, the protagonist used the pun 'ovary-annoying' to describe a woman who could not stop bragging about her having given birth to children. Here, we tackle a distinct but related issue - ovarian cancer. Recently, I had the privilege of reading an article published by the science blog of Cancer Research UK with regard to the re-evaluation of a present blood test for the diagnosis of ovarian cancer. [1] This re-evaluation yielded very encouraging results. In this article, I attempt to play the devil's advocate and explode certain myths regarding biochemical blood tests. 'Tumour markers', as we are told, are important for diagnosis. Is this the entire story though?
[2] Anatomy of Ovary - here we can see the developmental process of a follicle. Remember that the oocyte (egg cell) develops within the follicle at different stages of the female individual's life. After ovulation, the oocyte is pushed out of the follicle. The remaining cellular components form the corpus luteum (meaning yellow body in English; the original term is Latin). The corpus luteum is important for the production of female hormones such as oestrogen and progesterone. They are crucial in increasing the thickness of the endometrial wall (the middle layer of the uterine lining) in case there is a new 9-month visitor (aka the baby).
Before I start summarising the article and expressing my views, I have a disclaimer - I have not read the original research paper. I base my claims solely on the article published by Cancer Research UK (and later, the abstract of the study published by Cambridge University) for a number of reasons. Number one - if Cancer Research UK can promote the study clearly without resorting to distorting the data obtained from the study, then there is no need for me to consult the original paper to understand the foundation of the entire study. Number two - as Cancer Research UK uses this platform for the dissemination of health information to the general public, I, as a member of the general public, am already qualified to voice my opinions regarding how the information is presented and possibly identify some problems associated with this manner of presentation. I would like to utilise this platform to clear some of the misguidance on diagnostic tests. As I stated in my review of James Davies's 'Cracked: why psychiatry is doing more harm than good', I am adamant that biochemical tests are only 'informative'. Doctors are detectives (I still remember this on a slide presented in one of my pre-clinical pathology lectures, where the lecturer proceeds to say that she is, as a pathologist, the doctor of all doctors) and do not base clinical decision-making on one single test. Multiple factors are consulted before arriving at the relevant diagnosis.
Sonogram - Ovarian Cancer (extracted from https://spotlight.kaiserpermanente.org/wp-content/uploads/2018/10/Malignant_endometroid_OC_Picture3.png)
So, what does the article say? It highlights the significance of ovarian cancer as one of the most significant cancers around the world in terms of mortality and morbidity, where early diagnosis is crucial since survival rates are higher when treatment is provided.** There is a blood test used when determining whether a patient has ovarian cancer (Again, it's just one of the factors). It's called CA-125 and increases once there is irritation of one or more amidst a broad array of structures, including the epithelium of the Fallopian tubes, endometrium, endocervix, peritoneum (lining of abdominal cavity), pleura (the lining of the lungs) and pericardium (outer covering of the heart). [3-4] One of the models we use to determine whether or not to refer a patient to a gynaecology specialist is the risk of malignancy index (RMI), which comprises three main branches of information - CA-125 levels in the blood, ultrasound findings (there are different findings which warrant our attention, including signs of metastasis or ascites, i.e. fluid accumulation in the tummy, which can be indicative of tumour-induced inflammation), and the menopausal status of the patient. A score greater than 200 indicates higher risk and warrants specialist intervention. [5] The study that proposed the RMI model 30 years ago (yes, it was first published in 1990) by The London Hospital, employed a cohort from the hospital's gynaecology department. We call this 'secondary care', where the patients have already been referred by general practitioners (GPs). This is a potential problem since there is an intrinsic bias - of course there's a higher chance for these patients to have gynaecological issues, for they are referred and are likely to have symptoms serious enough to warrant that referral. A study performed by Cambridge University therefore wants to fill up the research gap and see about the blood test's utility in the primary care - amongst GPs. It is highlighted by Cancer Research UK that '10% of women with higher levels of the protein biomarker were diagnosed with ovarian cancer, making an abnormal test 12 times more predictive than previous estimates had suggested'. [1] It has also been stressed (though note the subtle inclusion of the term 'primary care') that over 380 women with an abnormal level of serum CA-125 had an alternative type of cancer, such as pancreatic cancer. It is also stated that age is an important factor, where 33% of women over the age of 50 with an abnormal test result had any cancer, as compared to 6% of those under the age of 50.
A table showing the parameters underlying RMI; extracted from: https://www.memorangapp.com/flashcards/34737/Clinical+24+Benign+Diseases+of+the+Ovary+%28Chapter+20+DS%29/ (As a note of clarification, Criteria A and B are graded while Criterion C is not. Giving a practical example, we now see a 60-year old post-menopausal patient. Under Criterion A, she gets 3 points. After doing an ultrasound, we discover that she's got multiloculated masses in both ovaries. She therefore gets a score of 3 under Criterion B. We do a CA-125 blood test and she reports a level of 50 U/mL. Multiplying all three figures (3 x 3 x 50 = 450) together, her RMI exceeds the 200 threshold. Therefore, she requires referral to a specialist.
Quoting the PI (principal investigator, the leader of the study), Dr Garth Funston:
'One of the most exciting things from this research is that we’ve been able to develop models which give the individualised risk or probability of a woman having cancer, based on their CA125 test level and their age.' [1]
Alright, so I lied. After reading the Cancer Research UK article, I still feel unsatisfied. As a bumblebee who loves reading (I used to read on my commute and continue until I reach the locker room), I've decided to read the paper selectively, focusing on the key features of the study that make Cancer Research UK so besotted. In general, I advise the exercise of caution and abstinence from jumping to conclusions. This is not only because of the fact that one study does not equate best medical practice (although CA-125 is recommended for patients with ovarian cancer symptoms in primary care, by NICE - National Institute for Health and Care Excellence), where systematic reviews are required to synthesise more reliable results, it is also because that the methodology of the study might have, on the outset, limited the applicability of its results.
The paper established different regression models to determine the relationships between serum (blood) CA-125 levels and the relative risks of ovarian cancer. [6] The models are done for different age groups. This broad range of age groups covers women of a variety of menopausal states. NICE Guidelines recommend that women with a 3% risk of ovarian cancer should seek urgent gynaecological referral, with a serum CA-125 level of 35 U/mL deemed significant enough to be a cut-off value. This study establishes different levels of CA-125 required for patients of different age groups to reach the 3% risk threshold.
[6] Regression Models extracted from the paper- It is actually not difficult to read this. Taking a patient of 30 years as an example. If this patient has a serum CA-125 level of 75 U/mL, the probability of having ovarian cancer will be nearly 4%. This exceeds the threshold of 3% set by national guidelines (UK). Therefore, the patient requires urgent gynaecological referral for additional tests and management.
This is in itself laudable, since serum CA-125 as a single factor can determine whether a symptomatic patient should receive advanced medical care. However, the study itself does have certain issues which are valuable for discussion. For purposes of clarity, I list them out in the following for reference:
Locality: This study utilises a retrospective cohort (meaning that it acquires data from databases and look at patient data) which recruits data from National Cancer Registration and Analysis Service, and Clinical Practice Research Datalink. The big problem is this - the diagnoses are restricted to those made in England only. Although the authors state that such data are significantly representative of the rest of the UK, I remain sceptical. This also means that further studies have to be performed to validate such results in a different country, due to intrinsic demographic differences;
Racial Differences: I am also quite shocked that, for the purposes of simplicity, the authors have regrettably ignored the importance of demographic factors. I sincerely understand their aim of synthesising a one-stop, easy-going test which solves the world's problems. However, I think in the long run, in order for this to work, we need to look at racial differences. Although it is an American study, it has reported that African American women have, given that other factors are controlled, generally lower serum CA-125 levels than their Caucasian (white) counterparts. [7] Sensitivity is a measure in medical research which assesses the chance of a patient being correctly diagnosed of a disease. In that study, the sensitivity of the serum CA-125 test (at the threshold designated by the American College of Obstetrics and Gynaecology) is 74.4% in Caucasian women. This drops to 33.3% in African American women. Other factors can be at play - however, this alone suggests that racial differences should be assessed. The UK is a racially diverse, highly globalised country and national medicine should be representative of all, not merely the majority.
Age Groups: The study records 50,780 entries in total. 31,086 of which are patients above the age of 50 years, whereas 19,694, under. It is a well-known fact that ovarian cancer occurs mostly in post-menopausal women. The median age of receiving an ovarian cancer diagnosis in the UK is 63 years, with the median age of women recruited in randomised controlled trials being 58 years. However, for women with predisposing genetic factors, they receive a diagnosis ten years earlier than the median age. [8] This fact aside, increasing age is also a risk factor for other malignancies. Results can be unfavourably skewed when we calculate incidence rates for all ages.
Possible confusion arising from benign disease: The authors have explained that a 1-year limit is applied. This means that the assumption that the cancer is already present when the test is done, all the way up to the point of diagnosis a year later, stands. The authors have also cited similar studies, saying that the 1-year limit is plausible, although they do not rule out the chance that incidental ovarian cancer(s) can arise after the CA-125 test. The most worrying scenario is when a benign ovarian disease has persisted throughout the 1-year limit and a non-ovarian cancer is diagnosed at the end of the period. Endometriosis, for instance, is a non-malignant entity which is significantly correlated with raised serum CA-125 levels. [9] In this case, it is all down to luck instead of science. The benign ovarian cancer may have no plausible correlation with the non-ovarian malignancy - their being two totally different entities having distinct pathophysiological mechanisms. Somehow, the serum CA-125 level is thought to be correlated with the non-ovarian lesion.
Different histological subtypes: this study mainly focuses on three categories of patients: no cancer, ovarian cancer and non-ovarian cancer; there is no further delineation of relationships between cancer risk and serum CA-125 level according to the histological subtype of disease. This ignores the fact that ovarian cancer is a very heterogenous disease, as reflected by the FIGO definition. [10] According to the results of the study, there is a positive correlation between the two. However, this is questionable. Although most ovarian cancers are epithelial [8], different histological subtypes bear different degrees of age-related prevalence. Younger patients, for instance, have more non-epithelial cancers such as malignant forms of teratomas and yolk sac tumours. These tumours may result in differing levels of rises of serum CA-125. Even the authors acknowledge that, more common in patients younger than 50 years, a greater proportion of invasive tumours like mucinous epithelial and non-epithelial cancers have lower chance of higher rises in serum CA-125. In other words, the serum CA-125 level of a younger patient may not adequately reflect the risk of ovarian cancer development. More importantly, what about patients with first-relative relatives who are diagnosed breast, endometrial or ovarian cancers? These cancers are related to specific gene mutations - BRCA1 and BRCA2. [11] They are also found in younger patients. Do BRCA-1/2-associated cancers produce differing levels of serum CA-125 as compared to other forms of ovarian cancer? If yes, this further muddles the waters.
Symptoms at a Glance: The focus of the study is the detection of ovarian cancers. This is a primary outcome of the study. However, the symptoms presented by the patients deviate, in terms of prevalence, considerably from those derived from an earlier study. [12] This study, published on JAMA, is performed in the US. The study aims to map the prevalence of different symptoms of ovarian cancer, in patients presenting to primary care. Back pain is the most common symptom, reaching 45%. Fatigue, in second place, reaches 34%. In third to fifth places, we have bloating (27%), constipation (24%) and abdominal pain (22%). However, in this study, 49.2% of patients present with abdominal pain. This is followed by abdominal distension/bloating at 23.4% and change in bowel habit (including constipation) at 11.8%. Fatigue only accounts for 7.0%. It has to be noted that all these figures are not mutually exclusive - meaning that patients can present with more than one symptom. Indeed, in the JAMA Study, the median number of symptoms presented by a patient is 2. [12] In my opinion, the discrepancies in initial clinical presentation do not necessarily have a significantly detrimental impact on the validity of the results. I think we can reconcile the two sets of the data without resorting to attributing them to geographical differences. It is well-known that ovarian cancers are called the 'silent killer' since symptoms are only obvious at advanced stages. [13] At advanced stages, such as stage 3 or 4, there are three possibilities: (a) the mass is large to an extent that it impinges on neighbouring structures, producing a broader range of symptoms. this can be illustrated by the diagram below.
Diagram extracted from Contemporary Ob/Gyn Net; It shows the female reproductive system in the pelvis.
As we can see, the ovaries are perched at the sides of the uterus. They are quite close to the ureters and colon. As the mass, at advanced stage, grows larger, there is greater risk of impingement of these structures. The sacral nerves, which arise at the sacral segment of the spinal cord, can also be impinged. This leads to back pain and abdominal pain. Some of the incidents of pain can be 'referred', meaning that pain is felt even when there is nothing wrong with that part of the body. A famous example is a patient with heart attack also has jaw pain.
(b) This can be due to the rise of nodal metastases. As the cancer spreads to the lymph nodes, this can lead to lymph node enlargement. The emergence of such masses leads to mass effect, compressing against neighbouring structures, giving rise to symptoms. Here is a diagram showing the intricate network of lymph nodes close to the female reproductive system:
Diagram showing pelvic and para-aortic lymph nodes (extracted from Macmillan Cancer Support).
(c) This can be due to the rise of multiple metastases. Ovarian cancers can spread to different parts of the body. They can spread regionally through peritoneal fluid to structures such as the colon, spleen and omentum (a layer of tissue which connects the stomach with other organs). The bladder can be affected. Moreover, in terms of distant metastases, the most commonly affected sites are the lung, liver and pleura. [14]
There is one more point I would like to add. In the Cancer Research UK article, it has emphasised that GP's should remain alert to the risk of other cancers when carrying out a serum CA-125 test. Disavowed by the authors of the paper, I also think this is a dangerous claim. Other cancers, such as colon cancer and gastric cancer, traditionally have more accurate tests for diagnosis. CA-125 can be incidentally high. More importantly, in the study, the sensitivity of CA-125 for ovarian cancer in all ages is 77% (the figure is calculated when CA-125 >= 35 U/mL, the guideline threshold), However, it drops dramatically to 29.1% for non-ovarian cancers. A CA-125 test can be done. However, its status as a marker for treatment follow-up and prognostication after diagnosis in a broad range of non-ovarian cancers cannot be ignored. Over-focusing on CA-125 during the diagnostic stage causes two major problems: (1) unnecessary patient distress, and (2) clinician getting blind-sighted. It goes without saying that even cirrhosis can lead to higher serum CA-125 levels, especially when there is ascites. [15]
Everyone likes a great success story. Everyone desires a brilliant medical breakthrough that can solve all our problems, especially when it comes to enigmatic diseases like cancer. However, critical evaluation and a healthy dose of scepticism are vital for the progression of contemporary medicine.
*Cover image extracted from: https://www.lalpathlabs.com/blog/wp-content/uploads/2017/02/ovarian-cancer.jpg. For possible, unidentified owners of previous cover images, if you wish to be acknowledged, feel free to contact me.
**Again, as a very sceptical person (who is even sceptical enough to check if I am using the word correctly with a dictionary), I do not like when people stress that early diagnosis equals early treatment and better prognosis in every single case. Of course, it is a maxim which rings true for most cases, but not all. For instance, in my blog post on prostate cancer, early diagnosis entails early intervention and early intervention entails higher risks in management. Prostate cancer is slow-growing and usually presents as an incidentaloma on imaging that probably won't bother the patient in most cases since it is mostly indolent. However, with the hype about getting rid of aberrations of the body just because they're there, patients risk an early death or impairment in quality of life attributable to the intervention itself. For instance, a prostate biopsy itself (Trans-rectal) can already lead to bleeding, infection and chronic prostatitis.
[1] Matson L. (2020). The CA125 blood test for ovarian cancer has been re-evaluated - Cancer Research UK - Science blog. Cancer Research UK - Science blog. Retrieved 22 November 2020, from https://scienceblog.cancerresearchuk.org/2020/10/28/an-existing-blood-test-for-ovarian-cancer-has-been-re-evaluated/.
[2] Utiger R. (2020). Ovary | animal and human. Encyclopædia Britannica. Retrieved 22 November 2020, from https://www.britannica.com/science/ovary-animal-and-human.
[3] Bischof P. (1993). What do we know about the origin of CA 125? European Journal of Obstetrics & Gynecology and Reproductive Biology, 49, 93-98.
[4] Sevinc A, Camci C, Turk HM, Buyukberber S. (2003). How to interpret serum CA 125 levels in patients with serosal involvement? A clinical dilemma. Oncology, 65(1), 1–6. https://doi.org/10.1159/000071198.
[5] Jacobs I, Oram D, Fairbanks J, et al. (1990). A risk of malignancy index incorporating CA 125, ultrasound and menopausal status for the accurate preoperative diagnosis of ovarian cancer. Br J Obstet Gynaecol. 97 (10): 922-9.
[6] Funston G, Hamilton W, Abel G, Crosbie EJ, Rous B, Walter FM. (2020). The diagnostic
performance of CA125 for the detection of ovarian and non-ovarian cancer in primary care: A
population-based cohort study. PLoS Med 17(10):e1003295. https://doi.org/10.1371/journal.
pmed.1003295.
[7] Dunton C, Bullock R, & Fritsche H. (2019). Ethnic disparity in clinical performance between multivariate index assay and CA125 in detection of ovarian malignancy. Future Oncology, 15(26), 3047-3051. https://doi.org/10.2217/fon-2019-0310.
[8] Jayson G, Kohn E, Kitchener H, Ledermann J. (2014). Ovarian cancer. The Lancet, 384(9951), 1376-1388. https://doi.org/10.1016/s0140-6736(13)62146-7.
[9] Karimi-Zarchi M, Dehshiri-Zadeh N, Sekhavat L, & Nosouhi F. (2016). Correlation of CA-125 serum level and clinico-pathological characteristic of patients with endometriosis. International journal of reproductive biomedicine, 14(11), 713–718.
[10] Prat J. (2013). Staging classification for cancer of the ovary, fallopian tube, and peritoneum. International Journal Of Gynecology & Obstetrics, 124(1), 1-5. https://doi.org/10.1016/j.ijgo.2013.10.001.
[11] de Jonge M, Ritterhouse L, de Kroon C, Vreeswijk M, Segal J, & Puranik R. et al. (2019). Germline BRCA-Associated Endometrial Carcinoma Is a Distinct Clinicopathologic Entity. Clinical Cancer Research, 25(24), 7517-7526. https://doi.org/10.1158/1078-0432.ccr-19-0848.
[12] Goff BA, Mandel LS, Melancon CH, Muntz HG. (2004). Frequency of Symptoms of Ovarian Cancer in Women Presenting to Primary Care Clinics. JAMA. 291(22):2705–2712. doi:10.1001/jama.291.22.2705.
[13] Jasen P. (2009). From the "silent killer" to the "whispering disease": ovarian cancer and the uses of metaphor. Medical history, 53(4), 489–512. https://doi.org/10.1017/s0025727300000521.
[14] Thomakos N, Diakosavvas M, Machairiotis N, Fasoulakis Z, Zarogoulidis P, & Rodolakis A. (2019). Rare Distant Metastatic Disease of Ovarian and Peritoneal Carcinomatosis: A Review of the Literature. Cancers, 11(8), 1044. https://doi.org/10.3390/cancers11081044.
[15] Zuckerman E, Lanir A, Sabo E, et al. (1999). Cancer antigen 125: a sensitive marker of ascites in patients with liver cirrhosis. The American journal of gastroenterology, 94(6), 1613–1618. https://doi.org/10.1111/j.1572-0241.1999.01152.x.