18FDG positron emission tomography versus 67Ga scintigraphy as prognostic test during chemotherapy for non-Hodgkin's lymphoma

Authors

  • Josée M. Zijlstra,

    1. 1 Department of Hematology , 2 Clinical PET Center , Departments of 3Clinical Epidemiology and Biostatistics, and 4Nuclear Medicine, VU University Medical Center, 5Department of Internal Medicine, Hospital Amstelveen, and 6Comprehensive Cancer Center, Amsterdam, The Netherlands
    Search for more papers by this author
  • 1,2 Otto S. Hoekstra,

    1. 1 Department of Hematology , 2 Clinical PET Center , Departments of 3Clinical Epidemiology and Biostatistics, and 4Nuclear Medicine, VU University Medical Center, 5Department of Internal Medicine, Hospital Amstelveen, and 6Comprehensive Cancer Center, Amsterdam, The Netherlands
    Search for more papers by this author
  • 2,3,4 Pieter G. H. M. Raijmakers,

    1. 1 Department of Hematology , 2 Clinical PET Center , Departments of 3Clinical Epidemiology and Biostatistics, and 4Nuclear Medicine, VU University Medical Center, 5Department of Internal Medicine, Hospital Amstelveen, and 6Comprehensive Cancer Center, Amsterdam, The Netherlands
    Search for more papers by this author
  • 4 Emile F. I. Comans,

    1. 1 Department of Hematology , 2 Clinical PET Center , Departments of 3Clinical Epidemiology and Biostatistics, and 4Nuclear Medicine, VU University Medical Center, 5Department of Internal Medicine, Hospital Amstelveen, and 6Comprehensive Cancer Center, Amsterdam, The Netherlands
    Search for more papers by this author
  • 2,4 Jacobus J.M. Van Der Hoeven,

    1. 1 Department of Hematology , 2 Clinical PET Center , Departments of 3Clinical Epidemiology and Biostatistics, and 4Nuclear Medicine, VU University Medical Center, 5Department of Internal Medicine, Hospital Amstelveen, and 6Comprehensive Cancer Center, Amsterdam, The Netherlands
    Search for more papers by this author
  • 5 Gerrit J. J. Teule,

    1. 1 Department of Hematology , 2 Clinical PET Center , Departments of 3Clinical Epidemiology and Biostatistics, and 4Nuclear Medicine, VU University Medical Center, 5Department of Internal Medicine, Hospital Amstelveen, and 6Comprehensive Cancer Center, Amsterdam, The Netherlands
    Search for more papers by this author
  • 4 A. Roelof Jonkhoff,

    1. 1 Department of Hematology , 2 Clinical PET Center , Departments of 3Clinical Epidemiology and Biostatistics, and 4Nuclear Medicine, VU University Medical Center, 5Department of Internal Medicine, Hospital Amstelveen, and 6Comprehensive Cancer Center, Amsterdam, The Netherlands
    Search for more papers by this author
  • 1 Harm V Tinteren,

    1. 1 Department of Hematology , 2 Clinical PET Center , Departments of 3Clinical Epidemiology and Biostatistics, and 4Nuclear Medicine, VU University Medical Center, 5Department of Internal Medicine, Hospital Amstelveen, and 6Comprehensive Cancer Center, Amsterdam, The Netherlands
    Search for more papers by this author
  • 6 Adriaan A. Lammertsma,

    1. 1 Department of Hematology , 2 Clinical PET Center , Departments of 3Clinical Epidemiology and Biostatistics, and 4Nuclear Medicine, VU University Medical Center, 5Department of Internal Medicine, Hospital Amstelveen, and 6Comprehensive Cancer Center, Amsterdam, The Netherlands
    Search for more papers by this author
  • and 2 Peter C. Huijgens 1

    1. 1 Department of Hematology , 2 Clinical PET Center , Departments of 3Clinical Epidemiology and Biostatistics, and 4Nuclear Medicine, VU University Medical Center, 5Department of Internal Medicine, Hospital Amstelveen, and 6Comprehensive Cancer Center, Amsterdam, The Netherlands
    Search for more papers by this author

Josée M. Zijlstra, VU University Medical Center, Brug 240, PO Box 7057, 1007 MB Amsterdam, The Netherlands. E-mail: j.zijlstra@vumc.nl

Abstract

Summary. A prospective study was performed, comparing gallium scintigraphy (67Ga) and positron emission tomography (PET) using fluorine-18 fluorodeoxyglucose (18FDG), to monitor the response of aggressive non-Hodgkin's lymphoma during treatment. 67Ga and 18FDG scans were performed in 26 patients after two cycles of CHOP (cyclophosphamide, doxorubicin, vincristine, prednisone) therapy. The scans were reviewed independently by four experienced nuclear physicians, who were blinded for the alternative scan technique and follow-up. Eleven out of 26 patients remained free from progression with a mean follow up of 25 ± 5 months, whereas 14 patients relapsed, and one died of lung cancer. Interobserver variation was significantly greater for 67Ga than for 18FDG PET. Some 64% of patients who had a negative early restaging 18FDG PET remained free from progression versus 50% of patients with negative 67Ga scans. Only 25% of patients with a positive PET remained disease free versus 42% of 67Ga-positive patients. Time to progression was associated with 18FDG PET results, but not with those by 67Ga. 18FDG PET had better test characteristics than 67Ga for the evaluation of early response in aggressive non-Hodgkin's lymphoma patients.

Less than half newly diagnosed patients with aggressive non-Hodgkin's lymphoma (NHL) are cured with CHOP (cyclophosphamide, doxorubicin, vincristine, prednisone)-like chemotherapy (Fisher et al, 1993; Janicek et al, 1997; Fisher, 2000). Patients who respond more rapidly to this front-line chemotherapy have a better and more durable response than slow-responding patients (Haw et al, 1994; Verdonck et al, 1995). Therefore, it is important to distinguish between patients who can be cured with standard approaches and those who may benefit from more intensive treatment.

During and soon after therapy, the conventional anatomically orientated radiological methods do not accurately predict patient outcome, mainly because of incomplete or delayed shrinkage of malignant tissues (Canellos, 1988; Surbone et al, 1988; Coiffier, 1999). Several studies have shown that computerized tomography (CT) scans during treatment do not differentiate patients who continued to have complete response at follow-up from those who relapsed (Verdonck et al, 1995; Front et al, 2000).

67Gallium citrate scintigraphy (67Ga) has been used to classify residual masses in patients after standard induction chemotherapy (Kaplan et al, 1990; Front & Israel, 1995). 67Ga binds to the transferrin receptor, which is abundantly present in rapidly growing tissues. The value of early restaging with 67Ga was demonstrated in a selected group of 30 patients with aggressive NHL and poor prognosis treated with high-dose chemotherapy (Janicek et al, 1997). Patients with a 67Ga-positive scan midway through chemotherapy had a much poorer outcome than those with a 67Ga-negative scan.

Recently, whole-body positron emission tomography (PET) using 18F-2-fluoro-2-deoxy-d-glucose (18FDG) has been introduced in the diagnosis and management of malignant lymphoma. Like 67Ga scintigraphy, it offers the possibility of differentiating viable tumour from inert residual masses (Front et al, 1992; Gasparini et al, 1998). Malignant tissue is characterized by an enhanced rate of glycolysis resulting from both increased expression of glucose transporter proteins at the tumour cell surface and increased hexokinase activity (Warburg, 1956). Compared with 67Ga, 18FDG has a more favourable biodistribution with much less uptake in liver, bone marrow, spleen and intestines. It is still unknown, however, which of these tracers has the highest avidity for lymphoma tissue in vivo. Furthermore, the performance of PET scanners in terms of spatial resolution and sensitivity clearly outweighs that of gamma cameras. On the other hand, 67Ga scintigraphy is more readily available in most centres and cheaper than PET.

If chemotherapy is effective, 18FDG uptake should be markedly diminished or even completely suppressed (Spaepen et al, 2001). The few studies that have addressed the role of 18FDG in defining early response during chemotherapy have suggested a role for PET in this setting by showing a decrease in 18FDG uptake before volume reduction on conventional imaging (Hoekstra et al, 1993; Romer et al, 1998; Mikhaeel et al, 2000; Spaepen et al, 2002).

To assess which nuclear imaging technique has the best characteristics as a response monitoring tool for aggressive NHL, a prospective study was performed comparing 18FDG PET and 67Ga scintigraphy after two cycles of CHOP chemotherapy.

Patients and methods

Patient population.  Between January 1998 and March 2000, 26 patients with histologically proven aggressive NHL were included in this prospective study. Patients were treated at the Department of Hematology, VU Medical Centre, Amsterdam, or at the Department of Hematology/Oncology, Hospital Amstelveen. Inclusion criteria were (1) diffuse large B-cell lymphoma, mantle cell lymphoma, follicle centre lymphoma grade III and anaplastic large cell lymphoma [revised European–American lymphoma (REAL) classification]; (2) measurable lesion; (3) no former treatment for NHL; (4) World Health Organization performance score 0–2; and (5) informed consent. According to the Declaration of Helsinki, the protocol was approved by our ethics committee.

Treatment and clinical follow-up.  Twenty-five patients were treated with standard CHOP that included 21-d cycles of cyclophosphamide 750 mg/m2 intravenously (i.v.), doxorubicin 50 mg/m2 i.v., vincristine 2 mg i.v. on d 1 and prednisone 100 mg orally on d 1–5. One patient received intensified CHOP containing cyclophosphamide 1000 mg/m2 i.v., doxorubicin 70 mg/m2 i.v., vincristine 2 mg i.v. on d 1 and prednisone 100 mg orally on d 1–5 every 14 d with the support of 300 µg of granulocyte colony-stimulating factor (G-CSF) subcutaneously (s.c.) from d 2–9. In 23 patients with stage IE or stage II–IV disease, the intended treatment was six to eight cycles of CHOP versus three cycles of CHOP followed by involved field radiotherapy in three patients with stage I disease.

Follow-up evaluation occurred every 3 months by a haematologist and included physical examination and a laboratory work-up. In case of new symptoms or palpable abnormalities, follow-up CT scans were performed. Patients were followed clinically for a minimum of 1 year or until death.

Radiographic evaluation.  Before therapy, all patients underwent baseline staging studies including standard histological examination of enlarged lymph nodes and bone marrow and CT scans of chest, abdomen and pelvis. Conventional or spiral CT was performed after the third cycle of CHOP and 4 weeks after treatment, respectively, after administration of i.v. contrast medium, using a Somatom Plus (Siemens) at the University Hospital in Amsterdam and a Tomoscan CX/S (Philips) in Amstelveen. The thickness of the transaxial slices of neck and thorax/abdomen was 5 and 10 mm respectively. Response to treatment was classified as complete response (CR), unconfirmed complete response (CRu), partial response (PR), no change (NC) or progressive disease (PD) according to standard criteria (Cheson et al, 1999).

18FDG PET and67Ga imaging.  Two weeks after the second CHOP cycle, 18FDG PET and 67Ga scans were performed. Clinicians remained blinded to the results of these scans to avoid any impact on patient management.

18FDG PET scans were performed using an ECAT HR+ scanner (Siemens/CTI, Knoxville, TN, USA). Patients were required to fast for at least 6 h before the scan. Serum glucose was measured and proved to be < 6·6 mmol/l in all cases. The scan trajectory comprised the inguinal to the cervical region. Emission scans (two-dimensional, 5 min/bed position) were started 60 min after i.v. administration of 350–420 MBq 18FDG, for a total imaging time of about 45 min. Images were reconstructed with an iterative reconstruction algorithm (osem, four iterations and 16 subiterations), resulting in a spatial resolution of ≈ 7 mm.

Immediately after PET, 185 MBq 67Ga citrate was administered i.v. After 72 h, scintigraphy (whole body) was performed using a dual-head gamma camera (ADAC Genesys, Milpitas, CA, USA), typically followed by single-photon computerized tomography (SPECT; chest and/or abdomen, n = 24). SPECT images were reconstructed with filtered back projection and a Hanning filter, resulting in a spatial resolution of ≈ 12 mm.

18FDG PET and67Ga assessment. 67Ga and 18FDG PET images were analysed visually by four experienced nuclear medicine physicians. The observers were blinded for clinical outcome, and each physician observed only a single scan modality for each patient. In a first session, they were also blinded for the baseline clinical staging findings. After an interval of at least 3 months, a second session was held in which the sites that were initially involved according to the routine clinical staging procedure were disclosed. Finally, for each patient and each modality, a consensus conclusion was formulated by both teams of observers, and this result was used in the analysis of test performance versus outcome.

In the first sessions, regions of abnormal uptake of 67Ga or 18FDG were scored as follows: 0 = benign, 1 = probably benign, 2 = unclear, 3 = probably malignant, 4 = malignant. In the final analysis, only abnormal focal tracer uptake was considered in sites clinically known to be involved at presentation. The observers were requested to formulate a final assessment in a consensus reading in one of three categories: positive, equivocal or negative for viable tumour.

Statistical analysis.  Interobserver agreement was measured with intraclass coefficients (ICC). Agreement between 67Ga and 18FDG PET consensus readings was measured with Cohen's kappa. Time to progression (TTP) was calculated from study entry to the first objective evidence of relapse or progression. Survival curves were estimated by the Kaplan–Meier method (Kaplan & Meier, 1958). Associations between scan results (67Ga and 18FDG respectively) and TTP were assessed using the log-rank test.

Results

Response to CHOP therapy

Twenty-six patients (14 male and 12 female) diagnosed with aggressive NHL and treated with CHOP were included. Their median age was 55 years (range 22–77 years). Patients' characteristics are listed in Table I.

Table I.  Patient characteristics (n = 26).
  1. REAL, revised European–American lymphoma classification.

Age (years)
 Median55
 Range22–77
Sex
 Male14
 Female12
Ann Arbor clinical stage
 I  5
 II11
 III  4
 IV  6
REAL
 Diffuse large B-cell20
 Mantle cell lymphoma  3
 Burkitt's lymphoma  1
 Follicle centre lymphoma  2

At standard mid-treatment evaluation (after three cycles of CHOP), CT and/or clinically defined response was considered insufficient (< 50% tumour regression on CT or progression after initial clinical response) in two patients, resulting in a switch to second-line chemotherapy and stem cell transplantation. One of them died after progression during second-line chemotherapy, and the other is still in complete remission (CR) 30 months after high-dose chemotherapy and stem cell transplantation (Fig 1).

Figure 1.

Clinical outcome after treatment with standard evaluation using CT scans. CR, complete response; CRu, unconfirmed complete response, PR, partial response; SD, stable disease; PD, progressive disease; NED, no evidence of relapse; RT, radiotherapy.

At conventional restaging after treatment, 15/26 patients (58%) achieved CR or CRu, nine of whom are still in CR with a median follow-up of 25 (range 14–34) months. One of the patients in CR after treatment died from lung cancer. Seven patients achieved partial remission (PR), and two progressed on therapy after first achieving transient PR. Four of the 15 complete responders and five of the seven partial responders relapsed within a median of 10 (range 3–17) months after completion of therapy.

Observer variability of 18FDG PET and 67Ga readings

Standard clinical staging before CHOP therapy revealed 61 sites involved in 26 patients. When observers were blinded to initial clinical staging findings, interobserver variability was significantly higher with 67Ga than with 18FDG PET [on a lesion basis: ICC 67Ga, 0·53, 95%confidence interval (CI) 0·22–0·72; 18FDG-PET, 0·98, 95%CI 0·97–0·99; on a patient basis the ICC was 0·67, 95%CI 0·28–0·85 versus 0·98, 95%CI 0·95–0·99 for 67Ga and 18FDG-PET respectively]. When the initial clinical staging data were made available, interobserver agreement increased for 67Ga (on a lesion basis: 0·80, 95%CI 0·66–0·88; on a patient basis: 0·74, 95%CI 0·43–0·88) and remained similar for PET.

18FDG PET and67Ga scans versus outcome

After two cycles of CHOP, 18FDG PET was positive in 12 patients and negative in 14 (Table II). Equivocal readings did not occur. In three patients, PET showed diffusely enhanced uptake at sites originally not affected at initial (CT) staging. According to the protocol, these scans were classified as negative, and the patients remained disease free at follow-up.

Table II.  Clinical outcome according to imaging results.
 PET+ Ga+PET+ GaPET Ga+PET GaTotal
  • *

    Including one patient with lung cancer.

Relapse6*31  414
No relapse213  612
Total8441026

67Ga scintigraphy was positive in 12 patients. One patient showed marginally enhanced uptake at an originally unaffected site, classified as negative and, in retrospect, follow-up did not show recurrent NHL. On a patient basis, results of 18FDG PET and 67Ga scintigraphy were concordant in 18/26 patients (69%).

18FDG-positive/67Ga-positive studies (n = 8)

Five out of eight patients showing accumulations in both 18FDG PET and 67Ga scintigraphy relapsed within 12 months. Relapse was proven by either biopsy (n = 2) or progressive disease on CT (n = 3). All relapses occurred in sites indicated by PET. One patient initially had cervical lymphadenopathy and a lesion in the right lung, which was also suspected to be lymphoma. After six cycles of CHOP, lymphadenopathy diminished, but the lung lesion increased. After lobectomy, this PET-positive site proved to be non-small-cell lung cancer. A few months later, the patient died of metastatic lung cancer, without evidence of relapsed NHL. Two patients with positive 18FDG PET and 67Ga scans after the second CHOP are still in continuous remission. However, one of them was treated with second-line therapy including stem cell transplantation because of insufficient response on CT at mid-treatment evaluation. The other patient is still in clinical remission 29 months after eight cycles of CHOP without further therapy.

18FDG-positive/67Ga-negative studies (n = 4)

In this group of four patients, three relapsed at 18FDG-positive sites, which were all extranodal (lung, liver and bones). The 67Ga scans did not show pathological 67Ga uptake at these sites (Fig 2). The remaining 18FDG-positive patient had diffusely enhanced uptake in the right lower lobe of the lung. His baseline CT scan had revealed a hilar mass and a consolidated right lower lobe. It was unclear whether this was pulmonary lymphoma or merely post-obstructive atelectasis with pneumonitis. The patient did not show signs of relapsed lymphoma during the follow-up period of more than 2 years.

Figure 2.

Prognostic value of 18FDG PET and 67Ga in a patient with presumed complete remission on CT. (A) Residual FDG uptake in the skeleton on PET. (B) A negative gallium scintigraphy with gallium uptake in the right lung was interpreted as a result of infection.

18FDG-negative/67Ga-positive studies (n = 4)

Out of four 18FDG-negative/67Ga-positive patients, one relapsed at the 67Ga-positive site, in a lymph node located near the renal pelvis. An 18FDG hot-spot in this area had been interpreted as physiological retention in the renal pyelum (Fig 3). The other three patients with a positive 67Ga scan are still in remission. These scans showed 67Ga avidity in regions that were clinically affected but also known as sites of physiological biodistribution of 67Ga (spleen, tonsil and parotid gland).

Figure 3.

Prognostic value of 18FDG PET and 67Ga in a patient with a negative PET (A) and a positive gallium scintigraphy (SPECT) (B). The FDG uptake near the right pyelum was interpreted as physiological retention near the right pyelum. The staging CT scan (C) showed a large mass medial from the right pyelum.

18FDG-negative/67Ga-negative studies (n = 10)

Six patients included in this subset remain in continuous clinical remission, whereas two experienced a relapse exclusively in the central nervous system. Retrospectively, an abnormal 18FDG uptake was seen in the cervical spinal cord 3 months before clinical relapse in one patient. Two other patients, both with a mantle cell lymphoma, relapsed 12 and 21 months after CHOP respectively.

At a median follow-up of 16 months (range 2–33 months), 64% of patients who had a negative early restaging 18FDG PET remained in clinical remission, whereas only 25% of patients with a positive PET scan remained disease free. However, 67Ga scans at that time point appeared to have no predictive value for treatment outcome (50% and 42% respectively).

The Kaplan–Meier curves for time to progression of patients with a negative or positive 18FDG PET or 67Ga scan are shown in Figs 4 and 5 respectively. The results of 18FDG PET after only two cycles of CHOP were significantly related to the time to progression after first-line treatment (log rank P-value = 0·05). 67Ga scan results did not show any relationship with the time to progression (log rank P-value = 0·67).

Figure 4.

Kaplan–Meier plot of progression-free survival in 12 patients with positive 18FDG PET compared with 14 patients with negative 18FDG PET after two CHOP cycles.

Figure 5.

Kaplan–Meier plot of progression-free survival in 12 patients with positive 67Ga compared with 14 patients with negative 67Ga after two CHOP cycles.

Discussion

The present study is, to our knowledge, the first head-to-head comparison of 67Ga scintigraphy and 18FDG PET in a cohort of NHL patients evaluated early during CHOP treatment. In this study, interobserver agreement and prognostic value of 18FDG PET were clearly better than those of 67Ga scintigraphy. Early recognition of patients who will fail to respond to first-line chemotherapy or who will relapse shortly after achieving a partial or complete remission is important, because these patients could be candidates for immediate, more intensive treatment rather than continuation of ineffective front-line chemotherapy with unnecessary toxicity. Mid-treatment CT scans do not discriminate accurately between patients who will achieve durable CR and those who will relapse (Front & Israel, 1995; Verdonck et al, 1995). Although the assessment of response by CT largely depends on the reduction in size of enlarged lymphadenopathy, functional imaging reflects the metabolic activity of tissues. Conceptually, this phenomenon may be more accurate in assessing treatment response than anatomical changes on CT. Furthermore, metabolic changes after therapy tend to precede anatomical changes and allow early response evaluation (Hoekstra et al, 1993; Dimitrakopoulou-Strauss et al, 1997).

Even though several other tracers have been proposed for use in malignant lymphoma (99mTc-MIBI, 201Tl, 111In-octreotide), the majority of the available data are for 67Ga and 18FDG. 67Ga imaging is useful for monitoring response or for detection of recurrence in lymphoma, especially above the diaphragm. Limitations are the low sensitivity and spatial resolution of gamma cameras and the physiological distribution of 67Ga. This may compromise abdominal evaluation and require additional imaging, which is inconvenient in settings where rapid decisions have to be made, as is the case with monitoring response in ongoing chemotherapy.

Evidence in support of 67Ga came from two recent studies. Janicek et al (1997) prospectively evaluated 30 patients with bulky advanced-stage aggressive NHL and poor prognosis with 67Ga scans at baseline and after two cycles of high-dose CHOP containing cyclophosphamide 4 g/m2 and doxorubicin 70 mg/m2. After a median follow-up of 31 months, 94% of patients with negative early restaging 67Ga scans remained free from progression versus only 18% of those with positive scans. However, these results were achieved in a highly selected group of patients treated with high-dose CHOP, which is not representative of all NHL patients treated with standard-dose CHOP. In another prospective study of 64 patients with aggressive NHL, a negative 67Ga after the first cycle of CHOP or CHOP-like chemotherapy predicted long-term continuous CR in 83% of patients. A positive 67Ga after one cycle predicted treatment failure in 64% of patients (Israel et al, 2002). The authors concluded that a positive 67Ga early during treatment might be used as an independent test in selecting patients who will not respond favourably to current treatment for early therapeutic modifications.

Despite the important role of 67Ga in lymphoma imaging, 18FDG PET might be a more effective tool, because of the inherent superiority of PET scanners over standard gamma cameras (Kostakoglu & Goldsmith, 2000) and the more favourable biodistribution of 18FDG compared with 67Ga. High avidity of 18FDG has been described for most types of lymphomas (Okada et al, 1991; Jerusalem et al, 1999). Finally, 18FDG PET is a single day procedure, and the radiation exposure is considerably less than with typical dosages of 67Ga. It has been known for more than a decade that changes in 18FDG uptake can be observed within days after the start of therapy (Hoekstra et al, 1993; Romer et al, 1998).

A few studies with smaller numbers of patients have shown that the evaluation of 18FDG uptake in lymphoma early during chemotherapy can predict the response to treatment. In a quantitative study, Romer et al (1998) reported that standard chemotherapy in 11 patients caused a rapid decrease in tumour 18FDG uptake as early as 7 d after treatment. The mean metabolic rates for 18FDG 7 d after initiation of chemotherapy were significantly lower in six out of 11 patients still in CR after a follow-up of 16 months. However, 18FDG uptake at 42 d was even better for predicting long-term outcome. In a heterogeneous study population of 28 patients with NHL, treated with different polychemotherapy regimens, Jerusalem et al (2000) reported that persistent 18FDG uptake after two to five cycles was predictive of CR, progression-free survival and overall survival. All five patients with, and seven out of 21 patients without, residual abnormal 18FDG uptake relapsed or progressed. In their study, the sensitivity of qualitative 18FDG PET imaging in identifying patients with a poor outcome was insufficient. Recently, Spaepen et al (2002) have assessed the value of mid-treatment 18FDG PET in predicting clinical outcome in 70 patients with aggressive NHL. The authors concluded that mid-treatment restaging 18FDG PET scans were highly predictive of progression-free and overall survival (P < 0·00001).

As visual analysis is the standard approach in this application, observer variability is an important issue, as has been shown for CT scanning (Fletcher et al, 1999). However, the interobserver variation was not thoroughly investigated for either tracer. In the present study, interobserver agreement was clearly better for 18FDG PET. As all observers had extensive experience with both techniques, this supports the notion that reading of 67Ga scans is relatively complex (Bar-Shalom et al, 2001). In part, this is explained by its biodistribution. In addition, it is likely that the image quality of 67Ga SPECT images versus 18FDG PET contributes to the observed differences. In this study, 67Ga doses were used as recommended in The Netherlands. It cannot be excluded that higher doses (e.g. 370 MBq) would have improved the test performance. The present data suggest that additional clinical data improve the interobserver performance with 67Ga.

In this study, scan data were compared with initial clinical staging. Additional baseline scans would have been useful but, before this study, it was anticipated that this would compromise patient compliance as neither procedure is standard for staging in our clinical setting. Moreover, 67Ga scintigraphy does not contribute to conventional clinical staging (van Amsterdam et al, 1996; Bonomo et al, 1997). This is not necessarily the case for 18FDG PET (Moog et al, 1998; Buchmann et al, 2000). However, in retrospect, no recurrence became manifest in clinically unsuspected areas designated as abnormal on 18FDG PET or 67Ga scans, except for a clinically unknown meningeal localization, which was retrospectively identified at 18FDG PET.

The primary aim of the present study was to compare two scintigraphic methods. Obviously, the sample size was too small to provide accurate estimates of predictive values. However, the results of 18FDG PET after only two cycles of CHOP were strongly related to the time to progression after first-line treatment (log rank P-value = 0·05). 18FDG PET had clearly better test characteristics than 67Ga scintigraphy in the evaluation of early response. More precise estimates of the predictive values of 67Ga scintigraphy and 18FDG PET could be obtained from a larger observational study. Alternatively, the impact of the application of either technique on patient outcome level would require a randomized study of 67Ga versus 18FDG PET in this context. The accumulated evidence on 18FDG PET and early response monitoring suggest that a randomized trial is now appropriate to determine whether the implementation of 18FDG PET in clinical management will improve patient outcome.

Ancillary