Age and Referral Route Impact the Access to Diagnosis for Women with Advanced Ovarian Cancer

Introduction

Epithelial ovarian cancer is the seventh most common cancer in women worldwide and the most lethal gynecological malignancy, with an overall five-year survival rate of approximately 40%.^1,2 The most important prognostic factor for the patient is the International Federation of Gynecology and Obstetrics (FIGO) stage at diagnosis, with a 70–90% five-year survival for early-stage disease (stage I and II). Unfortunately, the majority of women are diagnosed with advanced stage disease (stage III and IV), with a five-year survival of 20–40%.³ Epithelial ovarian cancer (EOC) consists of five histological subtypes, with high-grade serous carcinoma being the most common, accounting for 70%.⁴ EOC develops along two different tumorigenic pathways. Type I tumors are thought to arise from benign precursor lesions, and are often diagnosed at an early stage. Type II tumors include high-grade serous carcinoma and are more aggressive. These tumors develop from a precursor lesion in the fallopian tube and disseminate to the ovaries and peritoneum, initially giving vague symptoms and are often diagnosed at a late stage with poor survival.⁵ The most common genetic risk factors for EOC are BRCA1 and BRCA2 germline mutations, and are found in 15% of women with EOC.⁶

Most women with advanced ovarian cancer (AOC) do have symptoms prior to diagnosis, such as abdominal pain, bloating, urinary urgency, loss of appetite, increased abdominal size and change in bowel habits, sometimes several months before the diagnosis. The most common symptoms bringing women to the physician are abdominal pain, increased abdominal size, urinary urgency and loss of appetite. These symptoms are usually nonspecific and have a low positive predictive value.^7,8 If the symptoms are of recent onset, more frequent or severe, they are more likely to be caused by an ovarian tumor. Symptoms can occur at all stages of the disease and the majority of patients first present with symptoms to a general practitioner.⁸ However, a study from Denmark demonstrated that in only approximately 50% of cancer cases seen by a general practitioner were the patients’ symptoms interpreted as alarm symptoms. Therefore, symptoms not interpreted as alarm symptoms result in a longer diagnostic interval for the patient.⁹

Efforts have been made to speed up the diagnosis of ovarian cancer with screening programs using ultrasonography in combination with biomarkers in asymptomatic patients but without affecting the mortality rate in ovarian cancer.¹⁰ Over the last decade, several studies have investigated the cost effectiveness of screening programs for early detection and prevention of ovarian cancer, with mixed results.¹¹ Cancer patient pathways (CPPs) for AOC have been introduced in several European countries. The intention is to create a fast-track pathway for patients with suspected malignancy and to not take more time than necessary for diagnosis and treatment, and it is most often initiated in primary care. All CPPs act with criteria-based suspicion and fixed time duration for treatment.^12,13

Some studies have shown that reducing the time to diagnosis when the disease is already symptomatic does not affect overall survival in AOC or stage at diagnosis.^14,15 On the other hand, time intervals greater than 80 days have been shown to be associated with decreased survival.¹⁶ A study by Rai et al showed that guideline implementation with symptom triggered-testing, including CA125 and pelvic ultrasound, followed by urgent referral to rapid access clinics resulted in increased detection of ovarian cancers, but no stage shift was observed.¹⁷ Earlier diagnosis did not lead to a down shift in stage but could result in reduced cancer burden and facilitate complete surgical resection of the tumor.¹⁸ Since complete surgical resection is a major prognostic survival factor, it is important to speed up the process for each cancer patient.^19,20

In Sweden, CPP for ovarian cancer was implemented in 2016. The national guidelines include predefined alarm symptoms created by experts in the field. For ovarian cancer, the alarm symptoms were as follows: abdominal or pelvic pain, increased abdominal size or bloating, loss of appetite, frequent need to urinate, venous thromboembolism, abnormal or postmenopausal bleeding, respiratory problems or coughing, change in bowel habits or debut of inflammatory bowel syndrome.²¹

In an ideal flow, as shown in Figure 1, when a patient presents with alarm symptoms, the physician suspects ovarian cancer, and the patient is referred to a gynecologist for filter function. Both general practitioners, private gynecologists and other specialists refer patients to the filter function. The consultation includes a gynecological examination, vaginal ultrasonography using the International Ovarian Tumor Analysis group (IOTA) simple rules, CA125 level, and risk of malignancy index (RMI).^22,23 If the suspicion of ovarian cancer is rejected by the gynecological specialist, the CPP for ovarian cancer is closed. If reasonable suspicion of ovarian cancer remains, further investigation is performed, including computed tomography of the thorax, abdomen and pelvis and, optionally, a guided biopsy of the tumor, if it is a carcinoma of unknown origin and if primary surgery is not an option. The patient is then referred to a tertiary cancer center for a multidisciplinary cancer conference (MCC) to determine the final preferred mode of treatment. The guidelines in Sweden recommend all patients with AOC to be referred to a MCC at initial diagnosis. According to the Swedish national guidelines for ovarian cancer, the lead time in the CPP from suspicion of cancer by a gynecology specialist to the start of treatment is 24 days for upfront debulking surgery and 22 days for chemotherapy.²¹

Figure 1 Overview of the cancer patient pathway (CPP) for ovarian cancer in Sweden. Abbreviation: CA125, cancer antigen 125.

Aims

The aim of this study was to evaluate the timeline from the first presentation of symptoms to a physician until there is a reasonable suspicion of cancer among women diagnosed with advanced stage ovarian cancer. In addition, the most common symptoms for patients with AOC are documented. The possible role of age, mode of consultation, mode of appointment, CA125 level at time of diagnosis, histological subtype and place of residence were examined.

Materials and Methods

The health-care system in Sweden is publicly funded. Patients suspected of malignancy in primary care or regional hospitals are referred to a regional tertiary center. Skane University Hospital Lund Sweden is a tertiary center for all gynecological cancers and has a referral area for advanced ovarian cancer of approximately 1.9 million people.

This retrospective cohort study included patients diagnosed with AOC (FIGO stages III and IV) between January 1, 2017 and December 31, 2019 who were referred to Skane University Hospital Lund. All patients had a primary diagnosis with an International Statistical Classification of Diseases and Related Health Problems (ICD-10) code of C56.9 malignant neoplasm of unspecified ovary, C57.0 malignant neoplasm of fallopian tube, C76.2 malignant neoplasm of abdomen or C49.5 malignant neoplasm of connective and soft tissue of pelvis. Patients with relapse of AOC were not included in this study.

Data were collected from electronic medical records of Skane University Hospital (Melior.220-9.3.0.400–20210909.3, Cerner Corporation, Kansas City, USA). The following variables were recorded: age at diagnosis, time in days from first presentation with predefined symptoms to reasonable suspicion of cancer, first consultation in primary care or specialized care, first appointment as emergency or planned consultation, S-CA125 level at time before start of treatment, date of diagnosis, kind of symptoms for first consultation, place of patient residence, histological subtype and mode of treatment.

The group specialized care included both consultants in the hospital, regardless of specialty, and gynecologists in private health care. If a patient was referred from a general practitioner in primary care to the emergency department the same day as she first presented with symptoms, primary care was recorded as the first place of consultation. The diagnostic time interval was recorded as the time from the first physician consultation with predefined symptoms associated with ovarian cancer or incidental findings to the date when there was a reasonable suspicion of ovarian cancer.

The primary symptoms were classified as abdominal or/and pelvic pain, bloating or/and increased abdominal size, loss of appetite, urinary urgency, change in bowel habits, abnormal or/and postmenopausal bleeding, respiratory problems or/and coughing, venous thromboembolism and incidental findings. If more than a single symptom was presented, the symptom that led the patient to seek medical care was recorded as the main symptom. Only symptoms that were new within the last 13 months were included.

Patients’ residency at the time of diagnosis was recorded. According to the Swedish Association of Local Authorities and Regions, urban areas are classified as large cities and municipalities near large cities, with a population of at least 200,000 inhabitants. The urban area also includes medium-sized towns and municipalities with a population of at least 50,000 inhabitants and commuting municipalities near medium-sized towns. All other places are classified as rural areas.

Statistical Analyses

For descriptive statistics, continuous variables are presented as medians and first and third quartiles, and categorical variables are presented as frequencies and proportions (%).

Since the data were skewed, they were log transformed for the statistical analysis. A multivariable log-linear regression model was used to assess the relationship between time in days from first presentation with predefined symptoms to reasonable suspicion of cancer; one day was added as log (0) is undefined. The primary independent variable of interest was whether patients’ first level of consultation occurred in specialized care or in primary care. Other variables were age, type of consultation, mode of treatment, S-CA125 level and place of residence. All estimates were adjusted and presented as continuous variables with corresponding 95% confidence intervals (CIs), and presented in terms of interquartile ranges. The geometric mean was calculated for the time interval between the first presentation with symptoms and reasonable suspicion of cancer.

A simple nonparametric test as a complementary analysis was used when the distribution of the type of consultation was compared to the other variables, and an unpaired two-samples Wilcoxon test and the chi-square test of independence with Monte Carlo simulation were used. Data processing and statistical analysis were performed with the statistical software R.²⁴

Ethical Approval

The study complies with the Declaration of Helsinki and it was approved by the Swedish Ethical Review Authority (Dnr 2019-04450). Date of approval 2019-10-26. Informed consent was not required in this retrospective study. Our study is a non-intervention study with no direct medical implication on the patient’s health. Regarding GDPR and integrity questions, we secured the patients’ data, and the individual identification of the patient was disabled. All patients have a study number and all patient data is under locked deposition.

Results

A total of 252 patients were identified, and three patients were excluded from the cohort due to a lack of essential data (date of first consultation with predefined symptoms); thus, 249 patients were included in the study. The median age of the patients was 72 years. No differences were seen in median age in the patient group that first visited a general practitioner compared to those who visited a physician in specialized care.

Descriptive statistics are presented in Table 1. The majority of patients, 65.9% first presented with symptoms in primary care, whereas 34.1% had their first consultation in specialized care. Among the 249 patients included in this study, 53.4% were planned and 46.6% were emergency consultations. The median time from the first consultation with a physician to reasonable suspicion of cancer was 24 days, as shown in Table 2.

Table 1 Descriptive Statistics, Days from First Consultation to Reasonable Suspicion of Cancer per Patient Characteristics

Table 2 Patient Characteristics, Stratified by Level of Consultation

Figure 2 shows how different variables affect the time interval from first consultation to reasonable suspicion of cancer. Patient age was related to a longer delay to reasonable suspicion of cancer by 54.7% (16.6–105.3) 95% CI when comparing the first to the third quartile. When the first visit was in specialized care, a decrease in delay by 70% (−78.8 to −57.4) (95% CI) was observed compared to primary care. Emergency consultations were linked to a decrease in time delay by 52.2% (−65.9 to −32.9) (95% CI) compared to planned consultations. No significant difference in the time interval was found when comparing the different symptoms.

Figure 2 The variables (age, CA125 level, level of consultation, type of consultation, mode of treatment and place of residence) affecting the time interval between the first consultation and a reasonable suspicion of cancer. Abbreviation: CA125, cancer antigen 125.

Discussion

In this population-based retrospective study from Southern Sweden, approximately 66% of the patients diagnosed with AOC first went to a general practitioner to get their diagnosis. Nevertheless, the results showed that the median time interval from the first consultation to reasonable suspicion of cancer was significantly shorter when the first physician’s visit took place in specialized care compared with primary care, given other factors constant. Some patients might have presented directly to specialized health care as emergency visits if the waiting time to primary care was related to their symptoms. This study did not assess the grade of the symptoms at the first consultation, and there were no objective measurements of the symptoms. The overall time interval from the first consultation to reasonable suspicion of cancer was a median of 24 days, which is comparable to what has previously been reported for diagnostic time intervals for AOC.¹⁶ It has also been shown that long diagnostic delay among women with gynecological cancer could be caused by lack of cancer suspicion by referrals or a suspicion of other diseases.²⁵ This might be related to the relatively rare incidence of gynecologic cancer patients in general practitioners’ clinics. Compared to related cancers such as colorectal cancer, AOC is an uncommon disease and therefore not easily recognizable in everyday practice.²⁶

When the first consultation was an emergency visit compared to a planned physician’s appointment, the time interval to reasonable suspicion of cancer was significantly shorter. The reason for this could be that these patients had more severe symptoms, such as vomiting, severe abdominal pain or difficulty breathing, at the time of presentation, indicating a need for fast-track care. Moreover, others have shown that if the first consultation is at the emergency department, it is associated with a worse prognosis, which could be due to the patients having more advanced disease.¹⁶ It has also been shown that patients with poor survival had a shorter time to diagnosis.¹⁵ Similarly, in our study, the patients treated with the best supportive care had a rather short time to treatment, indicating that a poor performance score is a natural reason for faster referral.

Our results show that older patients had a significantly longer diagnostic time interval. This could be a result from misinterpretation of the patient’s status and attributing cancer symptoms to the normal aging process, which is understandable due to the uncommon diagnosis.⁸ Interestingly, the dominant symptom for women with AOC was abdominal or pelvic pain, which was present in almost half of the patients studied. On the other hand, many women presenting to primary care have vague symptoms that could be associated with harmless transient conditions but not with ovarian cancer.⁸ A recent study by Chan et al showed that more than 70% of patients with high-risk early-stage ovarian cancer have symptoms.²⁷

There is evidence that the way the physician interprets the patients’ symptoms, as vague or alarming, affects the lead time to diagnosis. If the physician suspects ovarian cancer, they are more often urgently referred to CPP and have a shorter diagnostic time interval. Patients with symptoms that are nonspecific or have comorbidities have longer diagnostic time intervals. A recently published study from Denmark showed that only one-third of patients with ovarian cancer were referred urgently to CPP.²⁸

In our study, patients with high CA125 values had a shorter time interval for CPP. This finding is, however, close to zero and therefore probably not important clinically. A recent study by Funston et al examined the performance of assessing CA125 levels in primary care and found that the test is useful for detecting ovarian cancer, particularly in women >50 years. Furthermore, an interesting finding was that nonovarian cancers could also be detected by the test, indicating that the CA125 test could help in decision-making for further investigation and urgent referral to a specialist.²⁹

When the place of residence was in a rural area, the median time interval was 22.5 days, and for urban areas, it was 28 days. There was no difference in the two groups, rural area or urban area, when comparing whether the first consultation was in primary care or specialized care. In our study, 68.7% of the women underwent surgery and received chemotherapy, which is the recommended mode of treatment.³⁰

A limitation of this retrospective study is that it was not possible to determine the severity of symptoms before the first consultation. We also do not know the date of symptom onset, which could have been months before the patient first presented to a physician. Patients might also have had their first contact with a nurse before their first visit to a physician. Another limitation is the small number of patients included in the study. Confounders such as performance status and comorbidity might have affected the results and thus the time from first consultation to reasonable suspicion of cancer. A strength of our study is that it is a population-based study and it includes all patients diagnosed with AOC in the population of Southern Sweden over three years in recent past.

Conclusion

In conclusion, we found that the time interval from first presentation with symptoms to reasonable suspicion of ovarian cancer was associated with whether the consultation was in primary care or specialized care, emergency or planned visit and the patient’s age. Improved awareness of ovarian cancer symptoms in primary care is of vital importance so that patients are referred urgently to gynecologists to reduce the time to diagnosis. Although no difference in overall survival has been shown for a longer time to diagnosis, it could be important at the individual level if more patients receive the recommended treatment of both surgery and chemotherapy. Educational programs for rapid admittance from health centers to gynecologists are warranted. Further research is needed to investigate whether other factors are associated with a longer time to diagnosis and whether confounders affect the results.