Division of Hematology and Oncology, Department of Medicine, Medical College of Wisconsin, Milwaukee, Wisconsin
Corresponding author: Timothy S. Fenske, MD, MS, Division of Hematology and Oncology, Department of Medicine, Medical College of Wisconsin, 9200 W. Wisconsin Avenue, Milwaukee, WI 53226; Fax: (414) 805-4617; email@example.com
Presented in part at the Annual Meeting of the American Society of Clinical Oncology; Chicago, Illinois; May 31 to June 4, 2013.
We thank the Donald J. Schuenke family for sponsoring Dr. Pingali's fellowship and the University of Nebraska Foundation along with the Nebraska Lymphoma Study Group for support of the lymphoma database at the University of Nebraska Medical Center.
The objective of this study was to compare the outcomes of patients with classical Hodgkin lymphoma (cHL) who achieved complete remission with frontline therapy and then underwent either clinical surveillance or routine surveillance imaging.
In total, 241 patients who were newly diagnosed with cHL between January 2000 and December 2010 at 3 participating tertiary care centers and achieved complete remission after first-line therapy were retrospectively analyzed. Of these, there were 174 patients in the routine surveillance imaging group and 67 patients in the clinical surveillance group, based on the intended mode of surveillance. In the routine surveillance imaging group, the intended plan of surveillance included computed tomography and/or positron emission tomography scans; whereas, in the clinical surveillance group, the intended plan of surveillance was clinical examination and laboratory studies, and scans were obtained only to evaluate concerning signs or symptoms. Baseline patient characteristics, prognostic features, treatment records, and outcomes were collected. The primary objective was to compare overall survival for patients in both groups. For secondary objectives, we compared the success of second-line therapy and estimated the costs of imaging for each group.
After 5 years of follow-up, the overall survival rate was 97% (95% confidence interval, 92%-99%) in the routine surveillance imaging group and 96% (95% confidence interval, 87%-99%) in the clinical surveillance group (P = .41). There were few relapses in each group, and all patients who relapsed in both groups achieved complete remission with second-line therapy. The charges associated with routine surveillance imaging were significantly higher than those for the clinical surveillance strategy, with no apparent clinical benefit.
An estimated 9290 individuals will be diagnosed with classical Hodgkin lymphoma (cHL) in the United States in 2013. Modern therapy is highly effective for cHL, with 5-year and 10-year overall survival (OS) rates of 83% and 78%, respectively, according to data from the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) Program. Routine surveillance imaging for patients in complete remission from cHL is common both in clinical practice and in clinical trials and is recommended in some treatment guidelines. Surveillance imaging offers the theoretical benefit of detecting asymptomatic relapses, which may allow for earlier implementation of second-line therapy and improved outcomes. Several studies have demonstrated that, in practice, surveillance imaging infrequently leads to the detection of asymptomatic relapse.[4-7] In addition, it has not been demonstrated that routine surveillance imaging actually improves survival or other major outcomes for patients with cHL in first remission.
In addition to an uncertain benefit, routine surveillance imaging has potential risks. Radiation exposure from surveillance scans can be significant and may increase the risk of secondary malignancies. By some estimates, radiation exposure from computed tomography (CT) scans may account for as much as 0.4% to 2% of cancers in the United States.[8, 9] Along with radiation exposure, routine surveillance imaging can lead to false-positive findings, unnecessary invasive procedures, significant patient anxiety, and a considerable increase in health care charges.[10-12]
To date, no studies have been published directly comparing outcomes of cHL patients managed with routine surveillance imaging versus clinical surveillance alone. Given the possible risks associated with routine surveillance imaging, combined with a lack of consistent evidence demonstrating a clear benefit, we undertook a comparison of outcomes of patients who underwent routine surveillance imaging versus clinical surveillance.
MATERIALS AND METHODS
Adult patients who were newly diagnosed with cHL between January 2000 and December 2010 at 3 participating academic tertiary care medical centers (Medical College of Wisconsin [Milwaukee, Wisconsin], University of Nebraska Medical Center [Omaha, Nebraska], and Washington University School of Medicine [St. Louis, Missouri]) were identified. To be included, patients must have achieved complete remission (confirmed by CT and/or positron emission tomography [PET] scan) at the end of first-line therapy and had to have ≥2 years of follow-up or have died <2 years after the end of treatment.
Pathology, clinical, and chemotherapy databases used at the 3 institutions were searched for patients with Hodgkin lymphoma. Patients were selected after screening for eligibility criteria. Charts were then retrospectively reviewed, and patients were classified into either the routine surveillance imaging group or the clinical surveillance group, based on the intended mode of surveillance. This was determined by the treating physician after the patient achieved first complete remission. In total, 241 patients met our inclusion criteria, with 174 patients in the routine surveillance imaging group and 67 patients in the clinical surveillance group. In the routine surveillance imaging group, the intended plan of surveillance was history, physical examination, laboratory studies, and surveillance CT and/or PET scans, which were ordered in advance of the follow-up clinic visit. In the clinical surveillance group, the intended plan of surveillance was history, physical examination, and laboratory studies, and scans were only obtained to evaluate concerning signs or symptoms that were noted during the clinic visit. The decision to follow the patient with clinical surveillance versus routine surveillance imaging was determined by the institutional policy or as dictated by the protocol for patients who were treated on clinical trials. Five patients were non-compliant with surveillance scans and, thus, were assigned to clinical surveillance on that basis.
In patients who had CT or PET scans available, the clinician notes and dates of scan orders were reviewed to determine whether the scans were preplanned versus ordered in response to symptoms or examination findings. If there was no documentation of symptoms or abnormal examination findings before the ordering of a scan, then that scan was considered to be a surveillance scan, particularly if the date the scan was ordered was far in advance of the actual date the scan was performed. If a patient had even 1 scan without having any signs or symptoms concerning for relapse, then they were placed in the routine surveillance imaging group. However, a scan that was obtained outside of the patient's lymphoma follow-up (eg, a CT scan ordered by an emergency medicine physician to evaluate for pulmonary embolus or renal calculi) was not considered a surveillance scan, because the intention was not to evaluate for asymptomatic relapse.
The primary objective of our study was to compare OS between patients who underwent routine surveillance imaging versus those who underwent clinical surveillance. Secondary objectives were to compare the rate of complete response to second-line therapy at first relapse in each group and to compare estimated imaging charges associated with the routine surveillance imaging and clinical surveillance strategies. OS was defined as the time from diagnosis of lymphoma to death. Time to relapse was defined as the time from first complete remission to the date of documented disease recurrence, based on either clinical or radiographic evidence.
Baseline patient-related and disease-related characteristics, including age, sex, disease stage, B symptoms, erythrocyte sedimentation rate (ESR), bulky disease status, and International Prognostic Factors Project (Hasenclever) score, were compared between the routine surveillance imaging and clinical surveillance groups. Age and ESR were compared using the Wilcoxon rank-sum test. Categorical variables were compared using the chi-square test, and exact P values were calculated for some results. Kaplan-Meier plots and log-rank tests were used to compare survival times, and cumulative incidence was used when competing risks occurred. Cumulative incidence rates were compared using a normal Z-statistic. Scan rates were compared using Poisson regression with deviance scaling and the natural log of follow-up time as an offset variable. The analysis was performed using the SAS statistical software package (version 9.3; SAS Institute Inc., Cary, NC).
Probabilities of OS were calculated using the Kaplan-Meier estimator. The 5-year survival probabilities with 95% confidence intervals (CIs) and P values for the point-wise test were calculated, and the log-rank test was used for overall comparison of survival curves.
Patient-Related, Disease-Related, and Treatment-Related Variables
Table 1 depicts the patient-related, disease-related, and treatment-related variables of the routine surveillance imaging and clinical surveillance groups. Mean age, sex distribution, and disease-specific characteristics such as stage, sedimentation rate, bulky disease, and international prognostic score were similar in each group. The number of patients who underwent radiographic surveillance versus clinical surveillance varied, depending on institutional policy. The University of Nebraska (n = 71) had a high proportion of clinical surveillance patients, with 62 in the clinical surveillance group and 9 in the radiographic surveillance group. The Medical College of Wisconsin (n = 54) had 1 patient in the clinical surveillance group and 53 in the radiographic surveillance group. Similarly, Washington University (n = 116) had 4 patients in the clinical surveillance group and 112 in the radiographic surveillance group. Frontline chemotherapy consisted of combined doxorubicin, bleomycin, vinblastine, and dacarbazine (ABVD) for 92% of patients in the routine surveillance imaging group versus 52% of patients in the clinical surveillance group. More patients in the clinical surveillance group received combined nitrogen mustard, doxorubicin, vinblastine, vincristine, belomycin, etoposide, and prednisone (the Stanford V regimen) (40% vs 3%). Primarily because of the more common use of the Stanford V regimen (for which radiation is a planned component), consolidative radiation therapy was used more frequently in the clinical surveillance group (75% vs 38%).
Table 1. Baseline Patient Characteristics and Treatment Received
OS was similar in both groups, with a 5-year OS rate of 97% (95% CI, 92%-99%) in the routine surveillance imaging group versus 96% (95% CI, 86%-99%) in the clinical surveillance group. The survival curves of the 2 groups did not differ significantly when they were compared using the log-rank test (P = .41) (Fig. 1).
Because the clinical surveillance group more frequently received the Stanford V regimen and consolidative radiation therapy, the OS of patients who received ABVD versus those who received Stanford V and for patients who did or did not receive consolidation with radiation therapy was compared, and there were no significant differences associated with either variable (Figs. 2 and 3).
Causes of Death
There were 5 deaths (2.9%) in the routine surveillance imaging group and 4 deaths (6%) in the clinical surveillance group. Overall, there was only 1 death from relapsed cHL, and that patient was in the routine surveillance imaging group. Of the 4 other deaths in the routine surveillance imaging group, 1 each was caused by congestive heart failure, cancer of unknown primary, complications related to hip fracture, and myelodysplastic syndrome. In the clinical surveillance group, there were no deaths from cHL. There were 2 deaths from non-Hodgkin lymphoma (1 diffuse large B-cell lymphoma and 1 peripheral T-cell lymphoma) and 2 deaths from unknown causes. It was confirmed that each of the 2 latter patients was in remission from lymphoma before death.
An analysis of the relapses is summarized in Table 2. There were 6 relapses in the routine surveillance imaging group and 5 in the clinical surveillance group. The 5-year incidence of relapse was 7.4% in the clinical surveillance group compared with 3.4% in the routine surveillance imaging group (P = .39). The average time to the detection of relapse was shorter in the routine surveillance imaging group at 18 months compared with 33 months in the clinical surveillance group. However, all patients who relapsed achieved complete remission with second-line therapy. All but 1 relapsed patient successfully underwent autologous hematopoietic cell transplantation. The 1 patient who did not undergo transplantation was not a candidate for the procedure because of comorbid conditions.
Frequency of Imaging and Estimated Charges of Imaging
Patients in the clinical surveillance group underwent an average of 0.21 CT or PET scans per year (95% CI, 0.18-0.31 scans per year), whereas patients in the routine surveillance imaging group underwent 0.98 scans per year (95% CI, 0.89-1.08 scans per year). This is a proportional difference, with a scan rate in the routine surveillance imaging group that is 4.5 times greater than the rate in the clinical surveillance group (95% CI, 3.1-5.5; P < .0001). The range of scans performed was 1 to 15 scans in the routine surveillance imaging group versus 0 to 5 scans in the clinical surveillance group. The number of scans performed per relapse detected was 14.6 scans in the clinical surveillance group compared with 127 scans in the routine surveillance imaging group. Assuming an average charge of $5190 per CT scan of the neck, chest, abdomen, and pelvis and $6600 per PET/CT scan, an additional charge of $17,700 was incurred per routine surveillance imaging patient. In terms of additional charge per relapse detected, $629,615 of additional estimated charges were incurred per relapse detected in the routine surveillance imaging group. This estimate does not include the charges associated with the evaluation of false-positive results identified on surveillance imaging studies.
Currently, there is no consensus regarding the optimal follow-up strategy for patients with cHL in first complete remission. Follow-up guidelines for routine surveillance imaging are based on expert opinion and vary widely. The most recent National Comprehensive Cancer Network guidelines recommend surveillance CT scans every 6 to 12 months for 2 or 3 years, whereas the European Society of Medical Oncology and International Working Group response criteria in lymphoma advise against routine surveillance imaging.[3, 17, 18] Although practice patterns vary, in the United States, patients with cHL (both on and off clinical trials) often undergo frequent and prolonged surveillance imaging, even for situations in which the risk of relapse statistically may be quite low, as in patients who have maintained first remission for 1 or 2 years or longer. This practice, although fairly common, occurs despite no demonstration of a survival benefit or other clinical benefit to routine surveillance imaging and also despite increasing concerns regarding the possible risk of secondary malignancies caused by radiation exposure from CT and PET scans. This latter risk is particularly relevant for young patients with cHL in first complete remission, because most of these patients will be cured of cHL and because the estimated risks of radiation-induced secondary malignancies are higher for younger patients.[19, 20]
Several previous studies demonstrated that, among patients with cHL who do relapse, the relapse most commonly is detected by the patient or the clinician and rarely is detected by routine surveillance imaging.[4-7] Such studies have documented that the detection of truly asymptomatic relapse with routine surveillance imaging is a rare event. Our study differs, in that outcomes were compared in patients who had cHL in first remission who were followed using clinical surveillance versus routine surveillance imaging strategies. By using this approach, in which patients from 3 tertiary care centers were assigned to the clinical surveillance group versus the routine surveillance imaging group based on the intended mode of surveillance, routine surveillance imaging was not associated with a survival benefit or with an improved outcome from second-line therapy. Because the number of relapses was very low, an analysis of differences in outcome between the 2 surveillance modes among relapsed patients was not feasible. It is noteworthy that the estimated charges for imaging (per relapse detected) were greater than $625,000 higher for the routine surveillance imaging approach. If a more “limited radiographic surveillance” approach were used (such as 3 surveillance CT scans over a 2-year period), assuming a 10% relapse rate, then we estimate that there would still be approximately $310,000 in additional imaging costs per relapse detected compared with clinical surveillance.
Our results are in agreement with a 2006 study by Guadagnolo et al, who used a Markov model to analyze the cost-effectiveness of CT for the routine follow-up of patients with HL in first remission. Those authors concluded that annual CT follow-up would be associated with a minimal survival benefit, including a decrement in quality-adjusted life expectancy for patients with early stage disease. For patients with advanced-stage disease, they concluded that annual CT follow-up would be associated with a very small gain in quality-adjusted life-years over non-CT follow-up, with an incremental cost-effectiveness ratio of greater than $9,000,000 per quality-adjusted life-year gained. Another recently published study estimated a cost of $12,608 per patient for surveillance imaging in individuals with supradiaphragmatic disease based on Medicare reimbursement. Our estimate of nearly $18,000 charged per patient in the routine surveillance group is in agreement with that study, because Medicare reimbursements typically are lower than the amount charged to third-party health insurance.
The reported failure-free survival rate for patients with cHL after frontline therapy with ABVD or Stanford V is 74% and 71%, respectively, when CT scanning was used to confirm complete remission. In the current study, the relapse rate was considerably lower because we selected patients with cHL who achieved complete remission with frontline therapy (determined by PET scans in >90% of patients in both arms). The 5-year relapse rate was 3.4% in the routine surveillance group and 7.4% in the clinical surveillance group, which are similar to recent studies in which PET-adapted therapy was used and event-free survival was in the 90% to 96% range.[24-28] A recent study that examined patients with cHL who achieved remission with frontline therapy and were followed by routine surveillance imaging versus clinical surveillance had relapse and overall survival rates similar those in our patients. Our study focused on patients who achieved complete remission (defined by PET scan) with frontline therapy. For higher risk patients, such as those who are not clearly in remission at the end of frontline therapy or possibly those who have a positive interim PET scan, more aggressive surveillance, including radiographic surveillance, may be appropriate.
It is noteworthy that, in both the clinical surveillance group and the routine surveillance imaging group, all patients achieved complete remission with second-line therapy. Also, the median time to relapse was 19 months in the routine surveillance imaging group versus 33 months in the clinical surveillance group. This suggests that routine surveillance imaging may allow for earlier detection of relapse and at an early stage in some patients who could be salvaged by radiation therapy alone. However, with the small numbers of relapses in both groups, we could not determine whether the opportunity of salvage by radiation therapy alone was lost. The finding that all patients achieved second complete remission argues against a critical advantage to detecting asymptomatic relapse and early initiation of second-line therapy. Thirty-five patients (20%) in the routine surveillance group and 5 patients (7.4%) in the clinical surveillance group had <3 years of follow-up. Given the low relapse rates in this patient population, combined with the tendency of relapses to occur in the first 1 or 2 years, longer follow-up of patients in either group is unlikely to yield different conclusions.
The strengths of this study include that it was a multicenter study with 241 patients, and approximately 25% of those patients were followed with clinical surveillance. We estimate that approximately 70% of all patients who received treatment for cHL (and 100% of patients with cHL who met eligibility criteria) from these 3 centers were included in this study. The patient-specific and disease-specific characteristics were very similar in the 2 groups, and only significant difference was that the clinical surveillance patients more commonly received Stanford V and radiation therapy. However, given the recently published results from Eastern Cooperative Oncology Group Study 2496, in which similar outcomes were reported for ABVD versus Stanford V, and because our main conclusion is that survival was comparable between the clinical surveillance and routine surveillance imaging groups, this difference in baseline characteristics is unlikely to bias our conclusions.
The main limitation of our study is that it is a retrospective study with the possibility that patients were selected for a certain surveillance mode based on clinical factors. However, because the decision to implement clinical surveillance versus radiographic surveillance was primarily determined by institutional policy, this type of bias is unlikely. Patients who received treatment at the University of Nebraska Medical Center were followed by clinical surveillance unless the patient was enrolled on a clinical trial. A small minority of patients from Washington University (3.4%) and the Medical College of Wisconsin (1.8%) did not undergo routine surveillance imaging because of patient noncompliance. All of the remaining patients at these 2 institutions underwent routine surveillance imaging. Also, we were unable to estimate whether the number of clinic visits was higher among patients who were followed by clinical surveillance or whether having a negative scan had an effect of reassurance in the routine surveillance group.
Our study has limited power to detect small differences in survival or relapse rates because of the small numbers of deaths and relapses that occurred. However, to perform an adequately powered prospective study to detect small (5%) differences in survival or relapse rates, a cohort of >1400 cHL patients would be required, which would be practically challenging. In a time of limited clinical trial resources, it seems unlikely that a prospective study of such scope will be undertaken to address this question.
Our results are in agreement with previous studies demonstrating that asymptomatic relapse (ie, relapse detected only by routine surveillance imaging) is relatively uncommon in patients who have cHL in first remission after frontline therapy.[4-7] Our analysis also is unable to demonstrate a significant survival benefit or improvement in outcome after second-line therapy in favor of routine surveillance imaging. In addition, our results highlight the considerable additional imaging charges associated with routine surveillance imaging. Patients in the routine surveillance imaging group underwent an average of 3 additional scan procedures. The additional radiation exposure from these extra scans would amount to approximately 75 milliSieverts (mSv) per routine surveillance imaging patient. This is equivalent to 27 years of background radiation exposure. By some estimates, this additional radiation exposure increases the risk of secondary malignancies in certain populations. For younger patients, like the population most affected by cHL, the lifetime cancer incidence may increase by as much as 1% to 2% because of exposure to 75 mSv of surveillance imaging radiation.[8, 9, 30-32]
Given the lack of apparent benefit from the extra imaging associated with routine surveillance imaging, combined with the significant additional costs and potential risks, clinical surveillance is not inferior to routine surveillance imaging. We recommend that patients who have cHL in complete remission with frontline therapy be followed with periodic history and physical examination and laboratory studies, with imaging reserved for those patients who have concerning signs or symptoms.
Dr. Pingali was supported by the Donald J. Schuenke Cancer Fellowship.
CONFLICT OF INTEREST DISCLOSURES
Dr. Bartlett serves on the Seattle Genetics Advisory Board. Dr. Armitage is a consultant to GlaxoSmithKline, Seattle Genetics, Genentech, Roche, Spectrum, and Ziopharm and serves on the Tesaro bio board of directors.