Earlier identification of chemoresistant disease, prior to treatment completion, would facilitate an individualized, risk-adapted strategy. Midtreatment (interim) metabolic imaging via FDG-PET has strong potential in this regard. In HL and aggressive B-cell non-Hodgkin's lymphoma (NHL), it has been repeatedly recognized that PET performed after only 2 or 3 cycles of chemotherapy provides valuable prognostic information. Early achievement of a negative PET (absence of abnormal FDG uptake concerning for tumor) is prognostically favorable, whereas persistent abnormal FDG uptake on early PET, even in the context of a CT response, raises concern for treatment failure [24
]. PET performed after only one cycle also appears to have prognostic significance [25
Why can interim PET provide more valuable prognostic information than one at the end of therapy? An early PET result may offer a window into the chemosensitivity of the tumor. As previously reviewed [21
], cancers are usually not diagnosed until they reach a size of 1010
cells. Chemotherapy kills tumor by first-order kinetics; that is, a given dose kills the same fraction, not the same number, of cells regardless of tumor size [27
]. Thus, in the idealized setting, each cycle of chemotherapy must kill approximately 1.5 to 2 logs of tumor for a lymphoma to be cured after 6 cycles. Accordingly, most of the tumor cell kill should occur early—with the first 2 cycles [21
]. A negative cycle 2 PET implies that the rate of tumor cell kill is sufficient to produce cure by the end of treatment, whereas a negative scan late in treatment does not distinguish between 4 logs of kill (PET remission) and the 10 or 11 logs of kill necessary for cure [21
As in NHL, the literature on the prognostic significance of interim PET in newly diagnosed HL represents a mixture of prospective and retrospective studies (), utilizing variable scanning techniques and criteria for PET interpretation [25
]. In a meta-analysis of interim PET in 360 patients with newly diagnosed advanced HL, the sensitivity of interim PET was 0.81 (95% confidence interval, 0.72 to 0.89) and the specificity was 0.97 (95% confidence interval, 0.94 to 0.99); corresponding sensitivity and specificity for diffuse large B-cell lymphoma were 0.78 and 0.87 [24
PET during first-line chemotherapy for Hodgkin's lymphoma.
FDG-PET performed after 2 cycles of ABVD (doxorubicin, bleomycin, vinblastine, dacarbazine) complements and is potentially superior to the IPS [39
]. In the widely cited study by Gallamini et al. [31
], 260 patients with poorer-risk HL were prospectively evaluated with cycle 2 PET, without change of therapy. The estimated 2-year progression-free survival was 95% in those with a negative interim PET, but only 13% in those with a positive interim PET. On univariate analysis, the cycle 2 PET result and the IPS were among the variables that were statistically significantly associated with treatment outcome. Notably, however, on multivariate analysis, only cycle 2 PET and the presence of stage IV disease were independently associated with outcome.
The majority of data on the prognostic significance of interim PET in HL is in patients with more advanced disease [31
]. Of these, the majority of patients had an IPS of <4 [24
]. The prognostic value of interim PET in early stage, favorable disease is not well established. One would expect, given the greater pretest probability of early stage patients being cured, that the high NPV observed in advanced HL would also hold for early stage HL provided that treatment is not de-escalated. In contrast to the high NPV of interim PET, the observed PPV has been variable (). Furthermore, relatively few patients with positive interim PET have been studied (50 in the largest series) [31
]. This is an important consideration when treatment intensification is contemplated on the basis of a positive PET, given the risk of greater toxicity with more intensive treatment.
A number of other factors may account for the observed variability in the PPV of interim PET. As in NHL, the variability is likely due in part to methodological differences among the studies. Variable (but mostly qualitative) criteria have been used in defining a positive scan, with borderline cases of “minimal residual uptake” [30
] posing particular challenges as later discussed. There are also differences in image acquisition protocols and technique; for example, earlier studies used PET alone, in contrast to combination PET/CT.
In addition, false positive signals from inflammation are a leading consideration [40
], as are other causes of false positive signals including infection, supraclavicular brown fat, granulocyte colony stimulating factors in bone marrow disease [41
], bony lesions, and in the case of posttherapy scans, thymic hyperplasia which can mimic mediastinal disease. False positives due to inflammation are possible in any type of tumor, but are a particular consideration in HL given its unique histopathology. In contrast to NHL, less than 1% of the tumor mass in HL is comprised of the malignant Hodgkin and Reed-Sternberg cells; rather, most of the tumor is comprised of nonmalignant cells, with the nature of that cellularity having prognostic significance [42
]. This raises the question, what are the histologic underpinnings of a positive PET signal in HL?
Moreover, the prognostic utility of any criterion can be a moving target, depending on the effectiveness of the underlying regimen. Thus, the prognostic and predictive value of interim PET could vary depending on whether one receives ABVD versus BEACOPP (bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine, prednisone) [39
], or chemotherapy alone versus combined modality therapy. The PPV of interim PET may indeed be lower after BEACOPP [36