We abstracted the clinicopathologic data from the cancer registry database of our department. Patients with disseminated cervical cancer were defined as patients who had clinically evident or biopsy-proven supraclavicular lymph node (SCLN) or hematogenous metastatic lesions. Patients with recurrent tumors were excluded. Patients with SCLN or hematogenous metastatic lesions that the physician believed were 'clinically evident' were included in this study, even if the patients did not receive a lesion biopsy. Patients with only paraaortic or mediastinal LNs metastasis on imaging were excluded as they did not indicate disseminated cervical cancer. Clinicopathologic data included age, histology type, number of involved sites, type of metastasis, type of primary treatment, treatment completion, chronic complications, and duration of survival after treatment.

The institutional review board approved this study (K-1001-002-036) and permitted us to waive informed consent.

Pretreatment computed tomography (CT) or magnetic resonance imaging (MR) of the abdomen, pelvis, and chest area was performed in all patients. In addition, positron emission tomography (PET) or PET/CT was performed in 22 of 30 patients. The number of involved sites and the type of metastasis were determined by the imaging findings. Specifically, metastatic lesions detected by CT or MR were regarded as positive lesions. LNs whose shortest diameter was longer than 1 cm were regarded as positive. Interpretation of lesions in the other organs was at the attending radiologist's discretion. However, lesions detected by only PET or PET/CT but not by CT or MR were regarded as negative because PET or PET/CT has high false positive rates.9 SCLNs that were enlarged or had fluorodeoxyglucose (FDG) uptake in the imaging studies were biopsied. The number of involved sites was the number of involved organs, except that LNs were further classified into pelvic, para-aortic, mediastinal, and supraclavicular LNs. When the involved sites outside of the pelvic organs were all LNs, the type of metastasis was regarded as lymphatic; otherwise, the type of metastasis was designated as hematogenous.

For 17 patients whose histological specimens were available, immunohistochemical staining for MMP-2, VEGF-A, and LAMC-2 were performed. MMP-2 (low vs. high) and VEGF-A (low, medium, or high) staining were graded qualitatively by a specialized pathologist who was blinded to the clinical data. LAMC-2 staining was graded as low if the percentage of stained cells was below 10%; otherwise, LAMC-2 was graded as high.

The survival curves were plotted with the Kaplan-Meier method. The association of overall survival (OS) with clinicopathologic variables and the expression levels of MMP-2, VEGF-A, and LAMC-2 were tested via the log-rank test. With variables associated with OS in log-rank test, we performed multivariate analysis using the Cox proportional hazard model. All statistical analysis was performed with SPSS ver. 14.0 (SPSS Inc., Chicago, IL, USA).

Between April, 1997, and May, 2008, 30 patients with disseminated cervical cancer were initially treated at our institute. The median age was 51 years and the most common histologic type was squamous cell carcinoma. Excluding the uterine cervix itself, pelvic and para-aortic LNs were the most commonly involved sites. The body of the uterus, bladder, mediastinal LNs, SCLNs, inguinal LNs, bone, lung, pancreas, liver, omentum, adrenal gland, peritoneum, and psoas muscle were listed as other involved sites.

Primary treatment was volume-directed radiation with concurrent chemotherapy (VDRCC), chemotherapy alone, or radiation alone in 17, 6, and 7 patients, respectively. VDRCC was defined as radiation to areas where tumors were radiologically present with concurrent chemotherapy. The chemotherapy regimens in VDRCC were as follows: cisplatin 75 mg/m² every three weeks in 10 patients, weekly cisplatin 40 mg/m² in 3 patients, and 5-fluorouracil (FU) 1,000 mg/m² for five days plus cisplatin 50 mg/m² every three weeks in 4 patients. In patients who underwent VDRCC, the radiation field included the whole pelvis area and other areas where lesions were identified by imaging. Specifically, radiation was administered to para-aortic LNs, mediastinal LNs, SCLNs, bone, and liver in 13, 3, 12, 1, and 1 patient, respectively. For pelvic radiation, the four-field box technique was employed. In some patients, a boost radiation to point B was administered at the discretion of attending radiation oncologists. The cumulative radiation dose to point A was between 81 and 91 Gy in 11 patients with vaginal brachytherapy. Radiation to areas other than the pelvis was performed simultaneously with pelvic radiation. Cumulative doses to para-aortic LNs, mediastinal LNs, SCLNs, and bone were 50-70 Gy, 45-56 Gy, 59-72 Gy, and 20 Gy, respectively. A patient received radiation to a hepatic lesion through image-guided stereotactic body radiation therapy using the cyberknife. In patients who received chemotherapy alone, the regimens were: paclitaxel plus cisplatin in 3 patients, 5-FU plus cisplatin in 2 patients, and topotecan plus cisplatin in 1 patient. Among seven patients who received radiation alone, 4 patients received only pelvic radiation, and 3 patients underwent radiation to para-aortic LNs and SCLNs in addition to pelvic radiation. Radiation to para-aortic LNs and SCLNs were performed simultaneously with pelvic radiation and cumulative doses were similar in patients who received VDRCC.

In 13 patients with lymphatic metastasis, the primary treatment was VDRCC, chemotherapy alone, or radiation alone in 9, 2, and 2 patients, respectively. In 17 patients with hematogenous metastasis, VDRCC, chemotherapy alone and radiation alone were performed in 8, 4, and 5 patients, respectively.

Among 30 patients, 5 could not finish the planned primary treatments and 4 patients experienced chronic complications of treatment. Namely, in the VDRCC group, one incomplete treatment, one esophageal stricture, and one rectovaginal fistula were observed. In the chemotherapy alone group, a subsequent radiation for persistent diseases after chemotherapy caused a rectovaginal fistula in a patient. In the radiation alone group, four incomplete treatment and one bowel obstruction from radiation colitis occurred (Table 1).

The median follow-up duration was 13 months (range, 1 to 140 months), and 21 deaths were observed. The median OS was 14 months and the 2-year OS rate was 36%. The 2-year OS rate of patients with hematogenous metastasis was 7%, but that of patients with lymphatic metastasis was 75%.

Immunohistochemical staining showed variable degrees of MMP-2, VEFG-A, LAMC-2 expressions (Fig. 1). The expression profiles of molecular markers are presented at Table 1.

In univariate and multivariate analysis, the type of metastasis and treatment completion were associated with OS (Table 1, Fig. 2). Patients with hematogenous metastasis had a 5.3-fold higher risk of death than those with lymphatic metastasis, and patients who could not finish the planned primary treatment had a 3.2-fold higher risk of death than those who finished the primary treatment (Table 2). However, none of the molecular markers showed association with OS.

Nine of 13 patients with lymphatic metastasis underwent VDRCC. Four of nine patients who underwent VDRCC had no evidence of disease (NED) at six months after primary treatment. Furthermore, three of four patients with NED at six months were disease-free at two years after primary treatment. However, all patients with hematogenous metastasis had persistent lesions at six months after primary treatment.