Impact of post-therapy positron emission tomography on prognostic stratification and surveillance after chemoradiotherapy for cervical cancer

Authors


Abstract

BACKGROUND:

A study was undertaken to investigate the detection of relapse and survival outcomes in patients with cervical cancer treated with curative intent chemoradiotherapy, and evaluated with a post-therapy 18F-fluorodeoxyglucose positron emission tomography (FDG-PET) scan.

METHODS:

Between January 2002 and June 2007, 105 consecutive patients were prospectively enrolled into a registry study designed to assess outcomes of chemoradiotherapy. A FDG-PET scan was performed between 3 and 12 months (median, 4.9 months) post-treatment at clinician discretion. Tumor response was graded as complete metabolic response, partial metabolic response, or progressive metabolic disease.

RESULTS:

Median follow-up was 36 months. At post-therapy FDG-PET, 73 (70%) patients had complete metabolic response, 10 (9%) had partial metabolic response, and 22 (21%) had progressive metabolic disease. Overall survival at 3 years was 77% in all patients, and 95% for those with complete metabolic response. On multivariate analysis, complete metabolic response (P < .0001) and pretreatment tumor volume (P = .041) were strong predictors for overall survival. The number of involved lymph nodes (P < .005) and International Federation of Gynecology and Obstetrics stage (P = .04) were predictive of relapse-free survival. In total, 18 patients relapsed at a single site, and 13 underwent salvage, with a 3-year survival of 67%. Patients with complete metabolic response had a distant failure rate 36-fold less than those with partial metabolic response (P < .0001). After complete metabolic response, only 1 patient (1.6%) relapsed without symptoms and was detected through physical examination.

CONCLUSIONS:

The presence of a complete metabolic response at post-therapy FDG-PET is a powerful predictor for survival after chemoradiation. The very low rate of recurrence in patients with a complete metabolic response justifies a conservative follow-up approach for these patients, because relapse is usually symptomatic and not detected by routine clinical review. Cancer 2011. © 2011 American Cancer Society.

Cervical cancer is the second most common cancer diagnosed in women worldwide, and is the fifth most common cancer overall.1 It accounts for approximately 150,000 deaths annually.2 Surgery is the primary treatment modality of early stage disease, but many women present with locally advanced disease and are optimally treated with combination chemoradiotherapy.3 Despite the burden of disease, there is no clear consensus among clinicians as to the optimal post-treatment surveillance.4 In a recent review of the literature,5 only 2 studies were identified that demonstrated a survival benefit as a result of post-treatment surveillance of patients with cervical cancers, both of which were retrospective.6, 7 Current follow-up practices include serial physical examinations with pelvic examinations, often in addition to cervical cytology (Papanicolaou test). However, routine surveillance with cervical cytology in these women has not been shown to be helpful for detecting recurrent disease.8 It has been observed that 59% of patients treated for gynecological cancer report increased anxiety before their hospital review,9 which may be attributable in part to the pelvic examination.

Our institution has previously reported a 5-year local control rate of 87% after curative intent chemoradiotherapy for patients with cervical cancer.10 Only 4% of patients had an isolated local failure at the time of their first relapse. We hypothesized that in the era of post-therapy 18F-fluorodeoxyglucose positron emission tomography (FDG-PET) scanning, of those patients who achieved a complete metabolic response, very few would recur with a local first site of recurrence. We predicted that in those patients who achieved a complete metabolic response, the clinical utility of pelvic examinations would be low. If our hypothesis is robust, it follows that a complete metabolic response may obviate the need for invasive serial pelvic examinations conducted in the hospital clinic setting. This study was designed to investigate the patterns of failure, survival, and salvage outcomes after curative chemoradiation of cervical cancer in relation to a single surveillance FDG-PET scan performed between 3 and 12 months post-therapy. The primary objective was to test the hypothesis that the asymptomatic relapse rate at 5 years (detected by clinical examination) would be less than the arbitrary value of 5%, a rate we considered clinically significant. Secondary objectives were to characterize pretreatment prognostic factors, patterns of failure (local, nodal, and distant), cause-specific survival, and overall survival.

MATERIALS AND METHODS

An independent human ethics board ratified this research. Informed consent was obtained in all patients. One hundred five consecutive patients from January 2002 to June 2007 were enrolled into a prospective tumor registry designed to assess outcomes to curative intent chemoradiotherapy. Patient characteristics (Table 1), staging, treatment, toxicities, and follow-up details were prospectively recorded. As the optimal timing of post-therapy surveillance FDG-PET was unknown, a single post-therapy FDG-PET scan was performed between 3 and 12 months at clinician discretion. Irradiation techniques have been previously described.11 During the study period, external beam radiotherapy was delivered to the pelvis to a dose of 40 grays (Gy) in 2 Gy fractions or 45 Gy in 1.8 Gy fractions. If pelvic nodes were involved, then nodes were boosted to a total dose of 50 to 50.4 Gy. Concurrent weekly cisplatin chemotherapy to a dose of 40 mg/m2 was administered during the external beam component of radiotherapy. Within 10 days of completion, a high-dose rate intracavitary brachytherapy boost was delivered twice weekly to a dose of 28 Gy in 4 fractions (or equivalent, to a total tumor dose of 80 Gy). All radiotherapy was completed within 7 weeks.

Table 1. Patient Characteristics
CharacteristicNo.%
  1. FIGO indicates International Federation of Gynecology and Obstetrics; SCC, squamous cell carcinoma.

Patients105100%
FIGO stage
 1b3533.3%
 II4845.7%
 III2221.0%
Histology
 SCC9287.6%
 Adenocarcinoma87.6%
 Other54.8%
Age, median y (range)54 (26-84) 
Tumor volume median mL (range)36.6 (0.1-628.3) 
Nodal status
 Node positive4341.0%
 Node negative6259.0%
No. of lymph nodes involved
 11413.3%
 21110.5%
 387.6%
 4+109.5%
Cause of death
 Cancer related2221.0%
 Other65.7%

All patients had histologically confirmed carcinoma of the uterine cervix, International Federation of Gynecology and Obstetrics (FIGO) stage Ib to IVa, and Eastern Cooperative Oncology Group (ECOG) performance status 2 or less. Staging included pelvic examination under anesthesia, whole body FDG-PET/computed tomography (CT), magnetic resonance imaging of the pelvis, and routine blood counts.

FDG-PET scans were performed using GE Discovery LS and GE Discovery STE dedicated PET/CT scanners with 4- and 8-slice multidetector CT, respectively (GE Medical Systems, Milwaukee, Wis). Patients fasted for at least 6 hours before intravenous injection of 370 MBq of fluorodeoxyglucose F-18. After the patient rested supine for 60 minutes, a PET/CT scan was acquired. Patients were routinely catheterized, hydrated, and diuresed. Metabolic changes post-therapy was scored as either complete metabolic response, partial metabolic response, or progressive metabolic disease.12

Clinical follow-up of patients including thorough medical history and physical examination was performed at 4 weeks post-therapy; every 3 months until 2 years post-therapy; every 6 months in years 3, 4, and 5; then yearly thereafter.

Statistical Methods

Times to events were measured from the date of the post-therapy PET. The asymptomatic relapse-free survival was characterized using the (nonparametric) Kaplan-Meier product limit method. The null hypothesis of a true asymptomatic relapse proportion of 0.05 was tested against the 2-sided alternative, and a P value for this hypothesis was calculated.

The association between metabolic response on post-therapy FDG-PET and each of several failure types (local, nodal, and distant) were tested using Cox proportional hazards models, using post-treatment metabolic response as a 3-level ordinal explanatory variable with values of complete metabolic response, partial metabolic response, or progressive metabolic disease. Likelihood ratio and Wald tests were used to test the significance of the association.

To characterize potential prognosticators for survival, a multifactor Cox proportional hazards model was built with overall survival as the response event. FIGO stage, tumor volume, number of pretreatment involved nodes, and the presence of complete metabolic response at post-treatment FDG-PET were used as candidate explanatory variables. Nonsignificant explanatory variables were eliminated from the model using stepwise backward elimination with a significance level of .05. The model was repeated to test for an association with relapse-free survival. To reduce bias, post-therapy metabolic response was excluded, as this variable was not independent of relapse. In the model for relapse-free survival, times to events were measured from the start date of radiotherapy. A noncancer-related death was considered a censoring event for cause-specific survival. The Kaplan-Meier product limit method was used to characterize the cause-specific survival and overall survival rates, and their respective 95% confidence intervals (CIs) for all patients, for the subsets of patients who achieved and those who did not achieve a complete metabolic response, and for those patients who had a single site of recurrence and subsequently underwent salvage based on the results of the post-therapy FDG-PET. Log-rank tests using chi-square statistic on 1 degree of freedom were used to test for equivalence.

RESULTS

The median time between the end of radiotherapy and the patient's post-therapy PET scan was 4.9 months. The PET was performed between 1 and 3 months in 23 patients, between 4 and 6 months in 53 patients, between 7 and 9 months in 26 patients, and between 10 and 12 months in 3 patients (Fig. 1). The median follow-up of living patients was 3.0 years (range, 0.8-6 years). At the close-out date of February 2009, 78 patients were alive, 25 patients were dead, and 2 were lost to follow-up. At the time of post-therapy FDG-PET, 73 (70%) patients had a complete metabolic response, 10 (9%) patients had a partial metabolic response, and 22 (21%) patients had progressive metabolic disease. A summary of findings at post-therapy FDG-PET is found in Table 2.

Figure 1.

Timing of post-therapy positron emission tomography and metabolic response are shown. CMR indicates complete metabolic response; PMR, partial metabolic response; PMD, progressive metabolic disease.

Table 2. Summary of Metabolically Active Disease at Post-Therapy FDG-PET
 Site of Disease at FDG-PET
StatusLocal OnlyNodal OnlyDistant OnlyLocal + NodalLocal + DistantNodal + DistantLocal + Nodal + Distant
  • FDG-PET = 18F-fluorodeoxyglucose positron emission tomography; P, 1 patient with palpable pelvic disease; N, 1 patient with palpable neck nodal disease.

  • a

    Salvage chemoradiotherapy performed.

  • b

    Salvage surgery performed.

Post-therapy disease41138231
Examination findingsPPP, NP
Salvage attempted9a1b
Alive3
Alive with disease52
Dead4336231

Rate of Asymptomatic Recurrences

Of the 73 patients with a complete metabolic response at post-therapy FDG-PET, 6 patients relapsed. The 5-year relapse rate for patients with a complete metabolic response was 0.097 (95% CI, 0.02-0.168). There was only 1 asymptomatic relapse detected by physical examination among the patients with a complete metabolic response. This patient relapsed at the primary site 1.65 years after the post-therapy PET. The estimated asymptomatic relapse rate at 5 years is 0.016 (95% CI, 0-0.046), significantly less than the nominated threshold rate of 0.05 (P < .03).

Patterns of Failure

At the time of the post-therapy FDG-PET, there were 96 patients who had no distant disease. Of these 96 patients, 10 went on to have distant failures. There was strong evidence of an association between post-therapy metabolic response and distant failure rate (P < .0001). Patients with a partial metabolic response had a hazard rate for distant failure 36 times (95% CI, 7-191) the hazard rate of patients with a complete metabolic response (P < .0001). On multifactor Cox proportional hazards model analysis, the pretreatment number of involved nodes (P < .005) and FIGO stage (P = .04) were both associated with relapse-free survival, whereas tumor volume was not. Patients with involved nodes had a 2.53-fold increase in risk of relapse per unit of time over patients without nodal involvement (95% CI, 1.32-4.86). On average, each additional positive lymph node increased the risk of relapse by a factor of 1.29 (95% CI, 1.17-1.42).

Five (7%) of 73 patients with a complete metabolic response and 1 (25%) of 4 patients with a partial metabolic response subsequently had a nodal relapse. There was evidence of an association between post-therapy metabolic response and nodal failure rate (P = .024). Patients with a partial metabolic response had a hazard rate for nodal failure that was 51 times (95% CI, 3-846) the rate of patients with a complete metabolic response (P = .0061). On multifactor Cox proportional hazards model analysis, complete metabolic response at post-therapy FDG-PET and pretreatment tumor volume remained significant variables after the process of eliminating nonsignificant variables.

Survival Outcomes

The overall survival rates for all patients at 3 years and 5 years were 0.77 (95% CI, 0.69-0.86) was 0.69 (95% CI, 0.57-0.82), respectively (Fig. 2). The cause-specific survival rates for all patients at 3 years and 5 years were 0.81 (95% CI, 0.73-0.90) and 0.72 (95% CI, 0.60-0.85), respectively (Fig. 3). There was a large survival difference noted between patients who achieved complete metabolic response at post-therapy FDG-PET and those who did not (see Figs. 4 and 5). For patients with a complete metabolic response, the 3-year and 5-year overall survival rates were 0.95 (95% CI, 0.89-1) and 0.91 (95% CI, 0.82-1), respectively, and the 3-year and 5-year cause-specific survival rates were 0.98 (95% CI, 0.93-1) and 0.93 (95% CI, 0.85-1). In contrast, patients without a complete metabolic response had a 3-year overall survival rate of 0.36 (95% CI, 0.21-0.60) and 3-year cause-specific survival rate of 0.40 (95% CI, 0.24-0.66) (Figs. 4 and 5).

Figure 2.

Kaplan-Meier product limit method for overall survival is shown for all patients with 95% confidence intervals.

Figure 3.

Kaplan-Meier product limit method for cause-specific survival is shown for all patients with 95% confidence intervals.

Figure 4.

Kaplan-Meier product limit method comparisons are shown for overall survival in patients with and without a complete metabolic response at post-therapy 18F-fluorodeoxyglucose positron emission tomography. CMR indicates complete metabolic response.

Figure 5.

Kaplan-Meier product limit method comparisons are shown for cause-specific survival in patients with and without a complete metabolic response at post-therapy fluorodeoxyglucose positron emission tomography. CMR indicates complete metabolic response.

Achieving a complete metabolic response at post-therapy FDG-PET was a strong independent prognostic factor for overall survival. Patients with complete metabolic response had a hazard rate 30-fold (95% CI, 9-105) less that of those patients without a complete metabolic response (P < .0001). Tumor volume was associated with overall survival (P = .046). Every 10-mL increase in tumor volume was associated with a 4% increase in rate of death. Forty-three (41.0%) patients had involved lymph nodes before treatment. Neither the pretreatment number of involved nodes nor FIGO stage was found to have an association with overall survival (P = .47 and P = .27 before elimination from the model). However, both the pretreatment number of involved nodes and FIGO stage were associated with relapse-free survival (P < .005 and P = .04, respectively). Patients with positive nodes had risk for relapse 2.53-fold (95% CI, 1.32-4.86) higher than lymph node-negative patients, with each additional positive lymph node increasing this risk by a factor of 1.29 (95% CI, 1.17-1.42).

Salvage Outcomes

Eighteen of the 32 patients who did not achieve a complete metabolic response had a single site of failure (local, nodal, or distant) at the post-therapy FDG-PET (Table 2). The estimated cause-specific survival rates at 1 and 2 years for this subset were 77% and 45%. Ten of these patients underwent salvage therapies (chemoradiotherapy in 9 and pelvic exenteration in 1). A further 3 patients had an initial complete metabolic response on post-therapy FDG-PET but later relapsed and underwent salvage therapy. Of the 3 who had an initial complete metabolic response, 2 recurrences were symptomatic, and 1 was asymptomatic. In total, the 13 patients undergoing salvage therapy had 1-year and 2-year overall survival rates of 89% and 67%, respectively, at a median follow-up of 1.6 years (range, 0.1-3.6 years). Eleven patients received salvage chemoradiotherapy for isolated nodal recurrences, and 2 patients received salvage surgery for isolated pelvic recurrences. Of the 3 patients who had an initial complete metabolic response and subsequently had an isolated failure amenable to salvage, 2 of the patients were successfully salvaged without further evidence of disease, including the patient who relapsed asymptomatically.

DISCUSSION

Our institution has previously reported a strong correlation with metabolic response and prognosis after chemoradiation in a variety of cancer sites, including nonsmall cell lung cancer,12 head and neck cancer,13 esophageal cancer,14 and rectal cancer.15 FDG-PET has been reported to be a powerful and highly sensitive imaging modality in the detection of primary and recurrent cervical cancer.16 In the post-therapy setting in cervical cancer, metabolic response on surveillance FDG-PET has been shown to prognosticate survival.17-19 Grigsby et al.20 reported a retrospective cohort of 152 women in which metabolic response at 3-month post-therapy FDG-PET was the most significant independent predictor of survival (P < .01). Women with no metabolic activity at the post-therapy FDG-PET had a 5-year cause-specific survival of 80% and a 5-year overall survival of 92%. Our series reports similar results, with a 5-year cause-specific survival of 98% and a 5-year overall survival of 91% in patients with a complete metabolic response. Patients without a complete metabolic response fared vastly worse (P < .0001); corresponding 3-year cause-specific and 3-year overall survival were 40% and 36%, respectively.

Our patterns of failure analysis also revealed that metabolic response was strongly associated with outcomes. There was evidence of strong association between post-therapy metabolic response with nodal failure rate (P = .024) and distant failure rate (P < .0001). Patients with a partial metabolic response had a relative risk of nodal failure 51 times that of patients with a complete metabolic response (P = .0061). Patients without a complete metabolic response at post-therapy FDG-PET were 36-fold more likely to develop metastatic disease (P < .0001) than those who did not have a complete metabolic response. When considering overall survival, the presence of a post-therapy complete metabolic response (P < .0001) and pretreatment tumor volume (P = .046) were favorable and unfavorable prognostic variables, respectively, whereas FIGO stage and number of involved nodes were not prognostic (P = .27 and P = .47, respectively). Similar to our previously reported cohort,10 FIGO stage and nodal involvement were strongly associated with relapse-free survival (P = .04 and P < .005). With respect to overall survival, it may be that the inclusion of the powerful prognostic factor of post-therapy metabolic response may have masked any association of the expected prognostic variables of FIGO stage and nodal involvement.

After curative treatment, only a small proportion of patients have a single site of failure suitable for salvage therapy. The majority of patients (70%) treated at our institution during this period achieved a complete metabolic response. Of the 32 (30%) patients who did not achieve a complete metabolic response, 9 underwent salvage chemoradiotherapy, 1 underwent salvage pelvic exenteration, and 22 were managed with palliative intent (Table 2). Pelvic examination did not add incremental benefit above FDG-PET; palpable disease was detected in 3 patients, all of whom had had multiple sites of metabolic disease not amenable to salvage. Twenty died from cancer, and 3 died from other causes. In total, 38 patients in our study failed treatment, with 18 patients presenting with a single site of failure (local, nodal, or distant), of whom 13 were suitable for salvage therapy. Of these 13 patients, 10 had isolated disease at the post-therapy FDG-PET, whereas 3 relapsed after an initial complete metabolic response. Historically, salvage hysterectomy after radiotherapy has been reported21 to show a 5-year actuarial survival rate of 49%. In the present study, 4 of 13 patients were successfully salvaged without further evidence of disease. Two of these 4 patients had an initial complete metabolic response at the post-therapy FDG-PET, 1 of whom was symptomatic at the time of relapse; the other was not and was detected through physical examination. Because of the low event rate, it is difficult to draw meaningful conclusions regarding salvage outcomes in patients with a complete metabolic response at a single post-therapy FDG-PET. Nevertheless, successful salvage could be achieved in patients presenting both symptomatically or asymptomatically.

The results of the post-therapy FDG-PET impacted highly on management decisions. Our previously published series of 249 patients suggests that selection of patients for salvage with conventional follow-up alone was unsatisfactory11; 15 salvage hysterectomies were performed, with only 5 having pathological confirmed disease and with only 2 remaining free of disease. Salvage hysterectomy in this series was associated with significant morbidity. Thirteen patients in the current report were included in that series, of whom none had salvage hysterectomies. On the basis of our previous results, we did not routinely offer salvage therapy during the early follow-up period unless isolated disease was demonstrated by the surveillance FDG-PET.

It is unclear whether early detection of recurrence translates to a survival benefit.22 There are no randomized prospective trials addressing survival benefit or cost-effectiveness of surveillance strategies. In a retrospective study of 583 consecutive patients, Morice et al.23 could not find a survival difference in those patients with recurrence detected symptomatically versus asymptomatically. Intensification of surveillance through serial FDG-PET examinations has recently been reported to have favorable outcomes in patients with early asymptomatic relapses.19, 24, 25 However, lead-time bias is a potential confounder, by which patients may appear to have an artificially longer survival from time of diagnosis of recurrence within the context of a shorter disease-free interval. Length-time bias, where less aggressive slow-growing disease is identified by the surveillance protocol and associated with better outcomes, is another potential confounder.

To consider whether serial physical (including pelvic) examinations could be omitted in patients who achieved a complete metabolic response at post-therapy FDG-PET, we tested whether asymptomatic relapses would be detected at a rate <5%, a threshold that we considered to be clinically significant. The majority (70%) of treated patients achieved a complete metabolic response. In this cohort, we detected a very low rate (1.6%) of asymptomatic relapses despite rigorous clinical follow-up. The findings of this study give strong evidence against the null hypothesis that the true asymptomatic relapse rate is equal to 5% (P = .03).

Our results indicate that post-therapy FDG-PET is a powerful prognostic tool after curative intent chemoradiotherapy of patients with cervical cancer. Recurrences after complete metabolic response at post-therapy FDG-PET are infrequent and very rarely detected through serial physical and pelvic examinations. Our findings suggest that patients who relapse after an initial complete metabolic response can be successfully salvaged when their recurrence is detected after the appearance of symptoms. In the context of the majority of patients who achieve a complete metabolic response at a single post-therapy FDG-PET (70%), the authors call into question the utility of follow-up with serial physical examinations. This study suggests that in this patient cohort, the exploration of less invasive and resource-intensive follow-up away from the hospital setting (such as telephone-based follow-up) without serial pelvic examinations may be warranted. A single post-therapy FDG PET may potentially be a cost-effective strategy affording greater convenience and less anxiety for patients, with reduced ongoing costs of outpatient services for healthcare providers.

The main limitation of this study is that it is a single-institutional study, and the results are not externally validated. However, our results are consistent with the survival data reported based on metabolic response that have previously been reported by Grigsby et al.20 and Schwarz et al.17 Another limitation of this study is that the optimal timing for the post-therapy FDG-PET cannot be concluded from our data. Patients in this study had FDG-PET scans at a median interval of 4.9 months after completion of therapy. In the setting of head and neck squamous cell carcinoma, there is evidence to suggest a high rate of false positives when a post-therapy FDG-PET is performed at 3 months.26 It is now our current practice to perform the post-therapy FDG-PET at 6 months after completion of irradiation. This strategy may reduce the likelihood of false positives by ensuring adequate time for resolution of metabolic activity, and may reduce false negatives by allowing time for population of micrometastatic disease to within the limits of spatial PET resolution. The disadvantage of this strategy is that early asymptomatic recurrences may be missed, but the impact of asymptomatic relapse detection for patient outcomes in this setting is as yet unknown. Further research with a well-designed case-controlled trial may aid in assessing the cost-effectiveness of post-therapy FDG-PET. In conclusion, a single post-therapy FDG-PET may obviate the need for serial pelvic examinations in the cohort of patients who achieve a complete metabolic response.

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

Ancillary