• computed tomography scan;
  • Hodgkin lymphoma;
  • positron emission tomography scan;
  • radiation exposure;
  • remission;
  • surveillance


  1. Top of page
  2. Abstract
  6. Acknowledgements


The majority of patients with Hodgkin lymphoma (HL) achieve disease remission after primary therapy. To the best of the authors' knowledge, no consensus exists for postremission surveillance imaging.


Retrospectively analyzed were 192 adult patients with classic HL in first remission. Events were defined as recurrent HL or secondary malignancies. Primary outcome was positive predictive value (PPV) of surveillance positron emission tomography/computed tomography (PET/CT) and CT scans in event detection. Secondary outcomes were costs and radiation exposures of surveillance scans.


Sixteen events (12 recurrent HL cases and 4 secondary malignancies) were detected during a median follow-up of 31 months. The PPV of surveillance PET/CT was 22.9% compared with 28.6% for CT (P = .73). Factors that were found to significantly improve the PPV of scans in detecting recurrent HL included PET and CT concordance, involvement of a prior disease site, or the occurrence of a radiographic abnormality within 12 months. There were too few events to determine whether event detection by PET/CT versus CT or the presence of symptoms at the time of event detection affected overall outcomes. The cost to detect a single event was approximately $100,000. Radiation exposure to detect a single event was 146.6 millisieverts per patient for each of 9 patients.


For patients with HL in first disease remission, surveillance radiography appears to be expensive, with limited clinical impact. Surveillance CT is generally adequate. Cancer 2010. © 2010 American Cancer Society.

Hodgkin lymphoma (HL) can be cured in >80% of patients, and in >90% with limited stage disease.1 The vast majority of patients achieve disease remission with multiagent chemotherapy with or without radiotherapy.2, 3 For patients with recurrent/refractory disease, salvage therapy confers long-term survivals of 25% to 90% depending on associated risk factors.4-6 To the best of our knowledge, there is no consensus regarding the optimal method of surveillance for the detection of recurrent HL among patients in remission. The 2009 guidelines from the National Comprehensive Cancer Network (NCCN) recommend routine surveillance imaging for patients in first remission (CR1).7 However, 3 retrospective analyses and a cost-effectiveness study argue against surveillance imaging in the absence of symptoms.8-10

Although the role of F-18-fluoro-deoxyglucose (FDG) positron emission tomography (PET) has been demonstrated in staging and in assessment of treatment responses for HL, particularly when performed in conjunction with computed tomography (CT) as part of a combined PET/CT scan,11-18 its role in the surveillance of HL patients in CR1 remains unclear. The NCCN discourages the use of surveillance PET scans because of anecdotal experience with false-positive radiographic findings.7 Two prospective series, however, suggest a possible role for PET scans in identifying early disease recurrence.19, 20

The risks of excessive surveillance radiography include high costs and radiation exposures, in addition to unnecessary procedures, follow-up scans, specialist referrals, and patient anxiety. Nonetheless, surveillance PET/CT and CT scans have been used routinely at our institutions since 2002. This retrospective study was designed to assess the utility of surveillance PET/CT and CT scans for HL patients in CR1.


  1. Top of page
  2. Abstract
  6. Acknowledgements


The diagnosis of classic HL was defined according to the World Health Organization (WHO) criteria,21 with review by dedicated hematopathology staff at Brigham and Women's Hospital (BWH) or Massachusetts General Hospital (MGH); central pathologic review was not performed for this study. All patients were staged according to the Cotswold modification of the Ann Arbor staging system.22 The end-of-treatment scan was defined as the first scan performed within 1 month after the completion of therapy. Definitions of complete remission (CR), partial remission (PR), stable disease (SD), and progressive disease (PD) were per the Cheson criteria.23 CR1 was defined either by the Cheson criteria for patients with an end-of-treatment scan or for patients without an end-of-treatment scan by an interim scan during therapy demonstrating a CR or PR followed by a first surveillance scan demonstrating no evidence of residual or active HL. The period of surveillance was defined as beginning either from the end-of-treatment scan or, for patients without an end-of-treatment scan, from the completion of therapy. Surveillance PET/CT or CT scans performed in response to a prior radiographic abnormality were defined as secondary surveillance scans; all other surveillance scans were defined as primary surveillance scans. The frequency of primary surveillance radiography was variable. In the first year, 30 patients underwent a single primary surveillance scan; 103 underwent 2 scans (or were scanned every 6-7 months), 34 underwent 3 scans (or were scanned every 4 months), and 25 underwent 4 scans (or were scanned at least every 3 months). In the second year, the majority of patients underwent primary surveillance scans every 6 months. The frequency of scans generally decreased thereafter.

An event was defined as a diagnosis of recurrent HL or secondary malignancy. A positive radiologic abnormality was defined as a radiographic finding leading to a change in medical management. A false-positive abnormality either resolved on further imaging within 6 months without treatment or was pathologically proven to be benign. A true-positive abnormality was clinically or pathologically demonstrated to be HL or a secondary malignancy. Recurrent HL was diagnosed either on the basis of tissue biopsy confirming the presence of recurrent disease or radiologic findings deemed to be consistent with lymphoma in the presence or absence of other local or constitutional symptoms. At the time of the identification of a positive radiographic abnormality, subsequent follow-up imaging studies were not counted as negative or positive, with the result that each case yielded a maximum of 1 positive radiologic abnormality, except in 2 cases (1 involving a PET/CT with anatomically distinct PET-positive [+]/CT+ and PET+/CT-negative [−] abnormalities and 1 involving a CT scan with biopsy-proven true-positive and false-positive lymph nodes). The positive predictive value (PPV) was defined as the number of true-positive results divided by the sum of true-positive and false-positive results.

Patient Selection

This study was approved by the joint institutional review board of Dana-Farber Cancer Institute (DFCI) and MGH; informed consent was waived. We retrospectively identified 345 records of HL from the cancer registries of DFCI and MGH from 2003 through 2006. The inclusion criteria were as follows: 1) treatment and surveillance at our institutions; 2) age ≥18 years; 3) histologic diagnosis of classic HL; 4) no history of other active malignancy; and 5) CR1. Of 345 records, 152 were excluded for the following reasons: consultation cases (n = 42), positive end-of-treatment imaging (n = 34), nodular lymphocyte predominant HL (n = 25), pediatric cases (n = 13), incomplete data (n = 12), treatment with noncurative intent (n = 9), concurrent malignancy (n = 6), duplicate records of patients seen at both institutions (n = 3), death during induction treatment (n = 2), incorrect histologic diagnosis, Epstein-Barr virus+ post-transplant lymphoproliferative disease (n = 1), incorrect year (n = 1), and unacceptable imaging modalities (n = 3).

PET/CT and CT Scans

Diagnostic CT scans included chest, abdominal, and pelvic imaging in most cases, and neck imaging in many instances. PET/CT scans performed at DFCI used a low-dose noncontrast CT acquisition for attenuation correction of the PET scan. PET/CT scans performed at MGH included a contrast-enhanced diagnostic CT (PET/CeCT) in addition to the standard low-dose attenuation correction CT. A total of 800 primary surveillance PET/CT (n = 474) and CT (n = 326) scans were performed during the follow-up period. Five CT scans were excluded for the following reasons: management plan pending (n = 3), sinus CT (n = 1), and radiologic diagnosis of appendicitis (n = 1). Among the 192 study patients, 83 had surveillance imaging exclusively with PET/CT, 27 exclusively with CT, and 82 with both PET/CT and CT. Data were collected from the reported radiologist interpretation of these scans; central radiology review was not performed. Designation of scans as true-positive or false-positive or false-negative was determined by radiology reports in conjunction with provider notes from clinic and hospital visits.

Primary Endpoint and Statistical Analyses

Cumulative survival probabilities were calculated using the Kaplan-Meier method. The primary endpoint for the study was the PPV of radiologic abnormalities detected on surveillance PET/CT and CT scans, as defined previously. Univariate analyses of PPV as a function of clinical and radiographic parameters were performed using the Fisher exact test.

Cost Analysis

To calculate costs of PET/CT and CT scans, we assumed that each PET/CT scan in our study represented either a whole-body or standard PET/CT, and that each CT scan represented an entire neck through pelvis survey. We used Current Procedural Terminology (CPT) codes for whole-body and standard PET/CT and neck, chest, abdomen, and pelvis CT scans to calculate Medicare reimbursements for hospital outpatient and nonfacility scans performed in Boston, Massachusetts.24

Radiation Exposure Analysis

Analysis of radiation exposures was restricted to MGH, in which standard PET/CT scans included CeCT as discussed previously. We calculated an average radiation exposure dose per scan based on data for 70 MGH patients in our cohort who underwent PET/CeCT scans and 58 patients who underwent CT scans. We compiled radiation exposure data for up to 4 PET/CeCT and 4 CT scans per patient, and then calculated the average radiation dose per scan per patient. Radiation from FDG-PET was recorded in millicuries (mCi) or megabecquerels (MBq) and converted to millisieverts (mSv) using the formula 0.019 mSv = 1 MBq.25 Radiation from PET/CeCT and CT scans was recorded as the dose-limiting product (DLP) in units of mGy−1⋅cm−1 and converted to mSv using the formula 0.017 mSv = 1 mGy⋅cm based on a 32-cm dosimetry phantom as an approximation for chest radiation exposure after chest CT.26


  1. Top of page
  2. Abstract
  6. Acknowledgements

Baseline Characteristics and Patient Outcomes

We identified 192 patients with classic HL in CR1 after the completion of initial therapy (Table 1). The median length of follow-up was 31 months (range, 6 - 66 months). Three patients died during the follow-up period, yielding a projected 4-year overall survival (OS) rate of 98.4% from the initiation of the surveillance period (Fig. 1). Sixteen events (12 cases of recurrent HL and 4 secondary malignancies) occurred, yielding projected 4-year progression-free and event-free survival rates of 93.8% and 91.7%, respectively.

Table 1. Baseline Characteristics of Study Population
  • ESR indicates erythrocyte sedimentation rate; WBC, white blood cell count.

  • a

    One patient had either nodular sclerosing or mixed cellularity histology. Another patient had either lymphocyte-rich or interfollicular histology. Nine patients had classic Hodgkin lymphoma, not otherwise specified.

No. of patients192
Age, y 
 Median (range)33 (18-81)
Duration of follow-up, mo 
 Median (range)31 (6-66)
 Male96 (50%)
 Female96 (50%)
 Nodular sclerosing155 (81%)
 Lymphocyte rich5 (2%)
 Lymphocyte depletion2 (1%)
 Mixed cellularity19 (10%)
 Othera11 (6%)
 Early (stage IA-IIA)92 (48%)
 Advanced (stage IIB-IV)100 (52%)
Early stage disease 
 Median age (range), y30.5 (18-77)
 Age >5013 (14%)
 B symptoms3 (3%)
 Bulky disease17 (18%)
 ESR ≥5013 of 61 (21%)
Advanced stage disease 
 Median age (range) y34.5 (18-81)
 WBC ≥ 15,000/mm322
 Age ≥45 y29
 Lymphocyte count <600/mm3 or <8% of WBC18
 Male gender55
 Albumin <4 g/dL65
 Stage IV disease31
 Hemoglobin <10.5 g/dL21
International Prognostic Score for advanced stage disease 
 Combined modality therapy (chemoradiation)82
 Chemotherapy alone110
 Chemotherapy regimens 
 Anthracycline-based chemotherapy189
thumbnail image

Figure 1. (Top) Overall, (Middle) progression-free, and (Bottom) event-free survivals of the study population are shown. In each curve, the y axis represents the proportion of the population in survival, and the x axis represents the length of time in months after the completion of primary therapy.

Download figure to PowerPoint

PPV of PET/CT and CT Scans

Among 795 primary surveillance PET/CT and CT scans, 7.8% (n = 62) were radiologically positive. Approximately 7.8% (37 of 474) of PET/CT and 3.1% (n = 10 of 321) of CT scans were designated as falsely positive (P = 0.009) (Table 2). The PPV of PET/CT scans in detecting recurrent HL or secondary malignancies was not statistically different from that of CT scans. PET/CT scans demonstrating concordant PET and CT findings (PET+/CT+) had a significantly higher PPV than those with discordant findings (PET+/CT- or PET-/CT+) (Table 3). In response to false-positive PET/CT and CT scans, physicians pursued secondary surveillance scans (n = 47), specialist referrals (n = 2), and invasive procedures for tissue diagnosis (n = 8).

Table 2. PPVs of Abnormal Findings on PET/CT and CT Scans in Detecting Recurrent HL or Secondary Malignancy
Imaging ModalityTrue-Positive ResultFalse-Positive ResultPPV, %P
  1. PPVs indicates positive predictive values; PET/CT, positron emission tomography/computed tomography; HL, Hodgkin lymphoma;.

Table 3. PPVs of Abnormal Findings on PET/CT Scans in Detecting Recurrent HL or Secondary Malignancy as a Function of PET and CT Concordance
Imaging ModalityTrue-Positive ResultFalse-Positive ResultPPV, %P
  1. PPVs indicates positive predictive values; PET/CT, positron emission tomography/computed tomography; HL, Hodgkin lymphoma; +, positive; −, negative.

PET+ CT+91242.9.02
PET+ CT−21810
PET- CT+070

All cases of recurrent HL were detected on scans performed ≤12 months after the initiation of surveillance, although a formal diagnosis of recurrent HL was not established until several months later in some cases. To further characterize positive radiologic findings, we examined the frequencies of positive PET/CT versus CT scans among patients with recurrent HL (9 vs 3, respectively) compared with those with no evidence of disease (25 vs 5, respectively) at 1 year after therapy. The difference between the ratios was not statistically significant (P = .67).

Detection of Recurrent HL

Twelve cases of recurrent HL were diagnosed during the surveillance period. Tissue diagnosis was obtained in 6, with the remainder diagnosed clinically on the basis of radiographic imaging being consistent with HL. Nine cases of recurrent HL were initially detected on PET/CT, and 3 were detected on CT. No major difference was noted in outcomes among recurrences detected by either modality. Of recurrences detected by PET/CT, 7 were initially detected on scans with PET and CT concordance (PET+/CT+), whereas 2 were detected on discordant (PET+/CT−) scans: 1 in a patient with recurrent bone disease, the other in a patient who eventually was found to have a mediastinal recurrence (and in whom a repeat PET/CT scan performed 3 months after the initial discordant [PET+/CT−] scan demonstrated concordance of PET and CT signals [PET+/CT+] in the mediastinum).

Influence of Symptoms on the Detection and Outcome of Patients With Recurrent HL

Seven cases of recurrent HL were accompanied by abnormal symptoms at the time of recurrence detection (6 of 8 patients with advanced stage disease, 1 of 4 patients with limited stage disease), whereas 5 were asymptomatic. Of the 7 patients with symptomatic recurrences, 3 were alive without disease, 2 were alive with disease, and 2 had died (1 from HL and 1 from graft versus host disease after allogeneic stem cell transplantation) at the time of last follow-up. Of the 5 patients with asymptomatic recurrences, 4 were alive without disease and 1 had died of HL at the time of last follow-up. The median time from the initiation of surveillance to a formal diagnosis of disease recurrence was 5 months for symptomatic cases (range, 3-10 months) and 7 months for asymptomatic cases (range, 4-15 months). It is interesting to note that in only 2 instances of recurrent HL were scans performed as a direct consequence of abnormal symptoms; at the time of last follow-up, 1 such patient was in CR after salvage therapy whereas the other was alive with persistent disease.

Factors Influencing PPV of Primary Surveillance Radiology Scans in Detecting Recurrent HL

We performed univariate analyses of clinical and radiographic parameters to determine which, if any, might influence the PPV of surveillance radiology scans in detecting recurrent HL (Table 4). We excluded cases of secondary malignancies for this analysis. For all surveillance scans (PET/CT and CT), the PPV in the detection of recurrent HL was found to be significantly increased for scans with an abnormality in a prior site of disease compared with scans without such an abnormality, and for scans with a positive radiologic finding at ≤12 months after the initiation of the surveillance period compared with scans at >12 months. The presence or absence of symptoms or mediastinal abnormalities did not appear to significantly influence PPV.

Table 4. Univariate Analyses of Effects of Clinical and Radiographic Parameters on PPV of Surveillance Radiology Scans in Detecting Recurrent HL
ParameterTrue-Positive ResultFalse-Positive ResultPPV, %P
  • PPV indicates positive predictive value; HL, Hodgkin lymphoma; PET/CT, positron emission tomography/computed tomography.

  • a

    For 1 patient with a false-positive PET/CT scan (positive PET/positive CT), the presence or absence of associated symptoms was not discernible on the basis of clinic notes.

For PET/CT and CT scans    
 Abnormality at previously involved site122235.3.0006
 No abnormality at previously involved site0250
 Radiologic abnormality at ≤12 mo123227.3.03
 Radiologic abnormality at >12 mo0150
 Mediastinal abnormality71433.3.09
 No mediastinal abnormality53313.2
 Associated symptoms presenta72621.2.99
 No associated symptoms presenta52020
 Associated symptoms present, advanced stage HL only61331.6.42
 No associated symptoms present, advanced stage HL only21214.3
For PET/CT scans only    
 Abnormality at previously involved site91833.3.006
 No abnormality at previously involved site0190
 Radiologic abnormality at ≤12 mo92526.5.09
 Radiologic abnormality at >12 mo0120
 Mediastinal abnormality61233.3.12
 No mediastinal abnormality32510.5
 Associated symptoms presenta62023.1.71
 No associated symptoms present31615.8

Detection of Secondary Malignancies

Four cases of secondary malignancies were diagnosed during the surveillance period: 1 lung fibrosarcoma, 1 paracaval leiomyosarcoma, 1 parotid basal cell carcinoma (BCC), and 1 acute myeloid leukemia (AML). A case of melanoma in situ was also diagnosed in a patient with recurrent HL, 13 months after the completion of salvage therapy. Two cases were detected by PET/CT and 1 by CT; the case of AML was initially suspected on the basis of physical examination showing bruising. All secondary malignancies were detected at ≥12 months from the initiation of the surveillance period and did not involve a prior site of disease. Two patients (1 with parotid BCC and 1 with AML) were symptomatic. After appropriate therapy, all patients were deemed to be in CR from their secondary malignancy at the time of last follow-up.

Cost of Primary Surveillance Scans

For MGH, a total of 148 primary surveillance PET/CeCT and 147 primary surveillance CT scans were performed, leading to the detection of 4 and 4 events, respectively (1 case of secondary AML was not counted as an event because detection was not based on surveillance radiography). For DFCI, a total of 326 primary surveillance PET/CT scans detected 7 events, and an additional 174 primary surveillance CT scans were performed that detected no events. We estimated that the cost in Medicare reimbursements of detecting 1 event among MGH patients was roughly $38,736.78 for detection by PET/CeCT and $61,820.48 for detection by CT. The cost of detecting 1 event among DFCI patients was $48,757.49 by PET/CT, in addition to $292,701.06 expended on CT scanning that failed to detect an event. For the entire population at both hospitals, approximately $111,349.07 was expended on primary surveillance PET/CT and CT scans to detect 1 event. This number is higher than that reported in an earlier cost analysis in Canada9 but more accurately reflects current practice in the northeastern United States.

Radiation Exposures From PET/CeCT and CT Surveillance Scans

We restricted our analysis of radiation exposures to MGH, at which the average radiation exposure was 39.65 mSv (95% confidence interval [95% CI], 37.7-41.6) per PET/CeCT scan, and 33.64 mSv (95% CI, 30.74-36.54) per CT scan. These numbers are comparable to published data from other series,27-29 although exposures are expected to be lower for PET/CT scans that do not incorporate CeCT scans (eg, scans performed at DFCI). Radiation exposure from surveillance PET/CeCT and CT scans at MGH during the follow-up period was approximately 146.6 mSv per patient, with 9 patients requiring surveillance imaging to detect 1 event.


  1. Top of page
  2. Abstract
  6. Acknowledgements

In our analysis of 192 HL patients in CR1 who underwent surveillance radiography with PET/CT and/or CT scans at our institutions, we observed the following: 1) PET/CT scans had significantly higher rates of false-positive results than CT scans; 2) PET/CT and CT scans had similarly low PPVs for the detection of recurrent HL or secondary malignancies; 3) PPVs of surveillance PET/CT and CT scans in detecting recurrent HL were significantly increased by PET and CT concordance, by involvement of a prior disease site, or by abnormal radiographic findings at ≤12 months after the initiation of surveillance; 4) just greater than half of recurrences were diagnosed in the presence of concurrent symptoms, although the presence or absence of abnormal symptoms did not appear to affect the PPV of surveillance scans; 5) there were too few cases of recurrent HL to adequately determine whether outcomes differed based on the presence or absence of abnormal symptoms, or based on detection by PET/CT versus CT; and 6) cost and radiation exposures incurred to detect a single recurrence or secondary malignancy were extraordinarily high.

In our cohort, either PET/CT or CT was adequate for the detection of recurrent HL. However, given the higher false positivity rate of FDG-PET, we favor surveillance imaging with CT, except in the specific anatomic case of bony involvement by HL, in which PET/CT appears to offer better detection. Abnormal CT scans should be followed by PET or PET/CT, particularly for findings in a prior site of disease. As a result of our analysis, our joint academic institutions have formally adopted the use of surveillance CT instead of PET/CT scans for HL patients in CR1 treated at our institutions. Data from the current study suggest that routinely scheduled surveillance imaging in the absence of symptoms might be curtailed or even abolished altogether, although further studies with larger numbers of patients and longer follow-up are needed to address this issue, perhaps in the setting of a prospective clinical trial. It is interesting to note that, whereas all disease recurrences in our cohort occurred within the first year after the completion of therapy, the literature supports the notion that the majority of HL recurrences occur within the first 3 years.2, 3, 30 This divergence in our observations compared with the literature may be partly explained by our designation of the starting time as the beginning of the surveillance period, rather than the date of the initial diagnosis or the initiation of first-line therapy; other potential contributing factors include variations in patient sampling, interpretation of radiologic or clinical findings, frequency of surveillance scans performed after the first year, and distribution of classic HL risk factors in our population (Table 1).

The current study has some limitations, including its retrospective design; the use of CeCT scans as part of the PET/CT acquisition in our radiation analysis, which may not reflect routine clinical practice in other centers; and the limited use of pathologic confirmation of radiologic positives. However, our retrospective design allowed for an actual assessment of cost expenditures and radiation exposures as reflected by real clinical practice. The length of our follow-up period validates our clinical and radiographic definition of true or false positivity with or without pathologic confirmation. Although we did not use a centralized review of radiology studies, such review was not essential within the parameters of our study because our interest was in provider response to radiographic interpretation at the time of imaging as opposed to third-party confirmation of radiologic interpretation.

To the best of our knowledge, the role of dedicated PET/CT in surveillance imaging of HL has not been examined previously in the literature. Several studies have explored the utility of surveillance PET scans in HL retrospectively.31-34 It is important to note that 2 studies have examined surveillance PET imaging in a prospective manner. A small study of 36 HL patients reported false-positive rates of 40% to 50% for surveillance PET20; notably, all recurrences detected on PET occurred before clinical suspicion of disease recurrence. The other, more recent, study enrolled patients with different lymphoma types, including 160 with HL19; the false positivity rate of PET was <1% overall but increased to 41.7% for PET scans that did not demonstrate a corresponding abnormality on follow-up CT. The low rate of false-positive findings was attributed to the investigators' strict definition of radiologic positivity, suggesting that the use of standardized criteria for positivity including standardized uptake values might allow for the judicious use of surveillance PET scans.35

To the best of our knowledge the cost of excessive PET-based surveillance imaging in HL has not been reported to date, whereas cost analyses for surveillance x-rays and CT scans have been published previously.9, 10 It is interesting to note that our calculations were restricted to primary surveillance scans and therefore did not reflect the actual costs incurred as a result of false-positive imaging studies, in terms of secondary radiology studies, follow-up procedures, biopsies, and additional clinic visits. Based on comprehensive modeling studies for surveillance CT scans incorporating these secondary costs,36 the total expenditures associated with false-positive surveillance PET/CT scans are expected to be higher than our estimates.

In the current study cohort, radiation exposure by PET/CeCT or CT scans to detect 1 event was approximately 146.6 mSv per patient for each of 9 patients. By contrast, the average radiation exposure for a standard chest x-ray is on the order of 0.01 to 0.02 mSv, although chest x-rays are no longer considered standard for the staging and surveillance of HL. To the best of our knowledge, the magnitude and significance of radiation exposures from surveillance radiography have not been previously established. However, a growing body of literature based on longitudinal studies in children and in atomic bomb survivors suggests that frequent radiology scanning at doses typically encountered with CT may increase the risk of malignancy, with 1.5% to 2% of all cancers in the United States possibly attributable to CT scans.28, 29, 37 In light of such unknown risks and high costs, we conclude that, for HL patients with a high remission rate, low risk of disease recurrence, and favorable prognosis, excessive surveillance radiography has little impact on overall outcome.


  1. Top of page
  2. Abstract
  6. Acknowledgements

We thank George P. Canellos, MD, and Susanna I. Lee, MD, PhD, for their critical reading of the article.


  1. Top of page
  2. Abstract
  6. Acknowledgements