In addition to effective screening and treatment of precursor lesions, another potential explanation for the reduction in cervical cancer mortality may be the improvement in therapeutic interventions. In this regard, the mainstays of treatment for cervical cancer, namely hysterectomy and radiation therapy, were introduced more than a century ago. These definitive management strategies have been expanded to include new ways of evaluating lymph node metastasis, such as computed tomography (CT), magnetic resonance imaging (MRI), and laparoscopic lymph node dissection, as well as less radical surgery for early-stage disease, and platinum-based chemoradiation. On the basis of 5 randomized phase III trials,^2-6 the National Cancer Institute (NCI) issued a Clinical Announcement in 1999,⁷ recommending that women who require radiation for the treatment of cervical cancer and their physicians consider the use of platinum-based chemoradiation. Thus, it seems logical to suggest that these recommendations would alter the practice of physicians and their decisions in regard to the definitive management of women with cervical cancer. This has not been examined in a population-based sample of patients treated in community settings in the US. To address the extent of this change in therapy, we undertook a population-based study evaluating patterns of care for women with cervical cancer, including data on how widely the use of chemoradiation has spread into community practice in the US.

In this study, staff at each participating SEER registry selected a sample of women who were newly diagnosed with histologically confirmed invasive cervical cancer in 1997, 2000, and 2001. For each diagnosis year, women aged ≥20 years at diagnosis were chosen at random within racial/ethnic groups. The SEER registries during 1997 and 2000 covered approximately 14% of the population in the US, and this increased to nearly 26% in 2001, by which time several new registries had joined SEER. During 1997, the metropolitan areas of Atlanta, Detroit, Los Angeles, San Francisco, San Jose, and Seattle and the states of New Mexico, Iowa, and Connecticut participated. In 2000, Utah and Hawaii also participated and in 2001, the states of New Jersey and Louisiana and the remainder of California joined. A total of 973, 773, and 1370 women with invasive cervical cancer were selected in 1997, 2000, and 2001, respectively. In 1997, non-Hispanic black and Hispanic women were oversampled. In 2000 and 2001, non-Hispanic black, Hispanic, Asian/Pacific Islanders, and American Indians/Alaskan Natives were oversampled. A case was not eligible if: 1) she had been diagnosed at autopsy; 2) the only evidence of her disease was based on a death certificate; 3) the cancer was not her first primary cancer (other than nonmelanoma skin cancer); or 4) she was simultaneously diagnosed with a second cancer, defined as within 2 months.

To ensure consistency and comparability of abstracting and coding, a centralized training of the primary medical records abstractor assigned to the study from each registry was conducted. Data from patient medical records were reabstracted to verify information on patient demographics, tumor characteristics, any treatment planned, given, or refused, and insurance status at the time of diagnosis. Each patient’s physician was contacted to verify treatment received or offered but refused by a patient. Follow-up for vital status is available through December 2004. Hospital bed size was collected, as was whether there was an approved residency training program at the facility where each patient received her most definitive treatment. Less than 1.5% of patients were treated at hospitals with an unknown residency training program; these were included in the category of no approved program. In addition, age, race/ ethnicity, insurance, International Federation of Gynecology and Obstetrics (FIGO) stage, grade, nodal status, receipt of treatments, year of diagnosis, and survival were examined. Women were grouped by insurance status: uninsured; private; any Medicaid; Medicare only; other and unknown insurance. Information on each patient’s comorbid conditions at the time of diagnosis was recorded from the hospital medical records. A single Registered Health Information Technologist coded these comorbidities centrally, and the Charlson comorbidity score was calculated from this information.⁹

The use of diagnostic exams and tests varied both by stage and by year of diagnosis (Table 2). About 45% to 50% of all patients with a known stage at diagnosis underwent an exam under anesthesia. This increased as a function of stage, with more women with stage II to IVA undergoing exams than women with FIGO I or IVB. Women who were unstaged were much less likely to have an exam under anesthesia. The most common imaging modality was CT, while MRI and ultrasound were used less frequently. The use of CT imaging was much lower in women with FIGO stage I than women with later-stage disease.

The treatment management differed substantially in women with FIGO stage I as compared with those with FIGO II-IVA disease (Table 3). Approximately 40% of women with FIGO stage I disease underwent a modified radical hysterectomy, while roughly 30% of women underwent total hysterectomy, and 15% local destruction. In 2001, less than 20% of women with stage II-IVA disease underwent hysterectomy.

Between 1997 and 2001, the use of both external beam (EBRT) and intracavitary radiation therapy (ICRT) increased (Table 3). For stage II to IVA, EBRT increased from 85% to 94%, and ICRT increased from 56% to 72% in the 4-year period from 1997 to 2001. This represents a marked change in the treatment management pattern for women with intermediate and locally advanced cervical malignancies.

There was a substantial increase in the use of chemotherapy in all stages between 1997 and 2001; this appears to be predominantly because of the use of combined chemoradiation (Table 3). For example, among women with stage II to IVA disease, the use of chemoradiation rose from 20% in 1997 to 72 % in 2001, as did the overall use of chemotherapy, which increased from 34% to 85% over the same time frame. A similar increase in both chemoradiation (28% to 59%) and chemotherapy (45% to 72%) was observed in women with stage IVB disease (Table 3).

Finally, in the models that examined the use of chemoradiation and the combination of chemotherapy, ICRT, and EBRT, use decreased with increasing age of the women (Table 4). Chemoradiation and the combination therapies increased with stage of disease and tumor grade. There was a significant increase in the use of concurrent chemoradiation between 1997 and 2000 and a significant increase in the use of combination chemotherapy, ICRT, and EBRT between 1997 and 2000, with a continued increase into 2001. There was no statistically significant association between either of these 2 treatment combinations and race/ethnicity or insurance status.

The most likely explanation for the change in therapy is the publication of 3 of the 5 clinical trials in the New England Journal of Medicine in 1999,^2-4 timed with the NCI-issued Clinical Announcement.⁷ Shortly thereafter, the fourth and fifth trials were published in the Journal of Clinical Oncology.^5,6 With the help of the American Society of Clinical Oncology, the American Society for Therapeutic Radiation Oncology, and the Society of Gynecologic Oncologists, the NCI Clinical Announcement was disseminated to those oncologists most likely to treat women with cervical cancer. A recently published population-based study reported by Barbera et al from the province of Ontario found a temporal impact of the NCI Clinical Announcement in Canada.¹⁰ In the Canadian study, the increase in the use of chemotherapy and radiation increased from <10% before the NCI Clinical Announcement to >67% in the period April 1999 through March 2001, after the announcement. This change was also reported in the patterns of care study by Eifel et al, where the use of chemoradiation for women with locally advanced disease rose from 19% in 1996 to 63% in 1999.¹¹ The increase in chemotherapy was not seen before 1999. The NCI Clinical Announcement was issued February 22, 1999. The reports by Eifel et al and Barbera et al suggest that these clinicians responded rapidly.

It is disturbing that only 70% of patients diagnosed with stage II to IVA disease received brachytherapy/ICRT as part of their treatment management plan. On the basis of patterns of care studies by Eifel et al,¹¹ we would expect that a larger percentage of women would receive ICRT. However, the lower percentage may reflect the finding that our data was population-based and included smaller community hospitals where, in the analysis by Eifel et al, patients were less likely to receive ICRT. The lower rates may also be because of several other factors, including lack of an appropriate facility to deliver ICRT in the community, distance to a facility offering ICRT, and patient compliance.

Finally, 11% and 5% of women with stage II to IVA disease did not receive radiation therapy in 2000 and 2001, respectively. Only 1% of patients in 2000 and 3.5% of patients in 2001 refused radiation therapy. We were not able to determine the explanation for this lack of therapy in the remaining patients.

Surgical therapies for women with FIGO stage I disease were 15% local destruction (including conization), 30% total hysterectomy, and 40% radical hysterectomy. These percentages were unchanged from 1997 to 2001. There was a much smaller increase in the use of chemotherapy and chemoradiation for women with early stage disease, and these results strongly suggest that surgical management remains the primary therapy for FIGO I cervical cancer.

In this study, we found that advancing age played a role in the therapeutic management of cervical cancer. Advancing age was associated with decreased receipt of concurrent chemoradiation and the combination of chemotherapy, EBRT, and ICRT. After adjusting for therapy, as well as other factors, the increased risk of all-causes mortality was higher in older women. However, there was no significant association between risk of death from cervical cancer and age after adjusting for race/ethnicity, insurance status, FIGO stage, grade, nodal status, treatment, and year of diagnosis. These results would suggest that older women, their families, and their healthcare providers need to be educated about the importance of screening for cervical cancer to find cervical cancer at an earlier stage. Furthermore, it may be essential to develop chemoradiation regimens that can be safely administered to older women with cervical cancer, particularly those with comorbidity.

At present there is no consensus in the developed world for the optimal pretreatment evaluation for cervical cancer. Despite efforts by FIGO to develop a standardized algorithm, physician- and institution-specific approaches remain in light of rapid technological developments in imaging modalities. Data from the joint ACRIN/GOG trial suggests that neither CT nor MRI provide reliable evaluation of cervical stromal involvement or lymph node involvement.¹² Institutions with expertise in positron emission tomography consider that particular modality a more accurate predictor of lymph node involvement, but there is still no national consensus on the optimal imaging modality.¹³ In addition, our findings clearly suggest that many clinicians remain convinced that exam under anesthesia, as compared with imaging, remains the most reliable means to assess of local tumor size and extent of disease.