Symptoms of Ovarian Cancer—Where to Set the Bar?

Ovarian cancer is the leading cause of death from a gynecologic malignancy among women in the United States and the fifth leading cause of cancer deaths among women overall after lung, breast, colorectal, and pancreatic.¹ Every year, approximately 23 000 women are diagnosed as having ovarian cancer and 14 000 women die of the disease.² One reason for the relatively high case-fatality rate is failure to identify early stage disease. While cure rates for early stage disease approach 90%, the overwhelming majority of women with ovarian cancer are diagnosed during stage III or stage IV, when cure rates are low. The failure to detect early stage disease has been attributed to inadequate screening tools and lack of early clinical symptoms.³

Attempts to identify sensitive and specific screening strategies for this disease to improve early detection have remained elusive. Ultrasonography, a screening tool that is relatively noninvasive, widely available, and low cost has been explored in a variety of settings. While several investigators have reported finding early stage lesions with transvaginal ultrasonography, the positive predictive value is too low to warrant its use in the general population.^{4 - 5} False-positive findings are particularly problematic in premenopausal women, in whom morphological features of the ovary vary considerably during the menstrual cycle. The tumor marker cancer antigen 125, routinely used in ovarian cancer patients to follow-up response to therapy, is fraught with high rates of false-positive and false-negative results in the screening setting.⁶ Ten years ago, a National Institutes of Health consensus conference on ovarian cancer concluded that there was no evidence to support routine screening in the general population.⁷ Several official groups have since reiterated this warning against routine screening with ultrasound or cancer antigen 125 for ovarian cancer.^{8 - 11}

In view of the failure to identify an acceptable screening approach for the early detection of ovarian cancer, symptom recognition assumes great importance. However, studies to identify a reliable pattern of symptoms of early ovarian cancer have been hampered by poor designs, reliance on retrospective recall, and small sample sizes. In studies that report the presence of symptoms prior to a diagnosis of ovarian cancer, the symptoms are often diffuse and nonspecific.^{12 - 13} The ensuing confusion has led to a polarization of positions, with patients and patient advocates claiming that their symptoms have been dismissed or overlooked on the one hand, and skeptical clinicians voicing reluctance to rely on reported symptoms to trigger an invasive diagnostic workup on the other hand.

In this issue of THE JOURNAL, Goff and colleagues¹⁴ compare self-reported symptoms of women prior to surgery for a pelvic mass with a group of control women presenting for care in primary care clinics. The authors found that women whose pelvic masses were found at surgery to be malignant had reported symptoms that were more severe, more frequent, and of more recent onset than those reported by the clinic population. As in other studies, the symptoms were of a broad nature and were referable to physicians seeing patients for the gastrointestinal tract, genitourinary, musculoskeletal, and general constitutional systems. Furthermore, all of these same symptoms were also reported to varying degrees by patients in the clinic population. Symptom reporting in the clinic population was correlated with age (premenopausal women reporting more symptoms, presumably due to menses, than postmenopausal women) and comorbidity (women with irritable bowel syndrome, diabetes, and thyroid disease reporting more symptoms).

The authors cite as a strength of their study the choice of controls, who were actively seeking medical care and therefore more representative of a general population of women in the health care system. However, the control population is on the average 10 years younger than the case population, and as symptom reporting is correlated with younger age, this discrepancy could tend to minimize the findings. Choosing women about to have surgery for a pelvic mass is an attempt to avoid the recall bias that has plagued previous studies, although it is possible that the anticipation of surgery for a potentially serious condition could also affect recall in the case population.

What are the lessons from this study? The findings are consistent with previous research,^{12 - 13 ,15} and lend strong support to the notion that ovarian cancer is often preceded by a set of recognizable symptoms. The study identifies the triad of bloating, change in abdominal size, and urinary symptoms, which occurred in 44% of the women diagnosed as having ovarian cancer. It also identifies 4 characteristics of the symptoms that should raise clinical suspicion of ovarian cancer, namely frequency, severity, time of onset, and total number of symptoms.

Unfortunately, the results do not provide clinicians the kind of decisional confidence that biochemical tests or definitive imaging studies provide in other settings. Symptom clusters must be interpreted in light of the age of the patient as well as the presence of other comorbid conditions. Given the relative prevalence and the diffuse nature of symptoms in the clinic population, it is still not clear where to set the bar for initiation of a diagnostic workup. For example, relying on the 4 characteristics of symptoms identified by the authors would still miss a considerable number of patients with a malignant lesion. Using a combination of 3 or more symptoms as a metric for considering ovarian cancer would have missed 56% of the women found to have a malignancy in this study. Similarly, using a severity score of 4 or greater for a symptom would have missed 29% to 91% of the affected women, depending on which symptom was reported.

How then do these findings help an individual patient and her clinician interpret the potential seriousness of her symptoms, particularly at a time when many in the medical profession have begun to lose faith in clinical skills and have adopted a narrowly focused biotechnological approach to patients? Based on these data, there is no way to avoid the conclusion that early diagnosis of ovarian cancer must rely on the elusive practice of clinical judgment, a skill that involves careful analysis of the characteristics of the presenting symptoms set within the context of a thoughtful dialogue between the patient and her physician. The importance of this study is not the validation of a symptom cluster as a precise way to diagnose ovarian cancer, but rather the reinforcement of the need for an ongoing process of communication between patients and their physicians. In the absence of more definitive diagnostic tools, early detection of ovarian cancer will continue to challenge both the artistic skill of astute clinicians as well as their accumulated scientific acumen.