Interim and end-treatment positron emission tomography scan in aggressive B-cell lymphoma: we still lack an interpretation key

In this issue of the journal, Mylam et al. [1] relate to the impact of interim and end-of-treatment positron emission tomography (iPET and ePET) interpretation by the treating physicians in patients with diffuse large B-cell lymphoma (DLBCL). In this study iPET was performed after 2–4 cycles of chemotherapy, and ePET 2–16 weeks after completion of treatment in a cohort of 430 patients with DLBCL, treated with a bi- or three-weekly CHOP/CHOEP (cyclophosphamide, doxorubicin, vincristine, prednisone/cyclophosphamide, doxorubicin, vincristine, etoposide, prednisone) program. Treatment was given at eight Danish specialized centers of hematology during a time period of 4 years, between 2005 and 2009, with a mean follow-up time exceeding 3 years (3.4 years). A total of 617 PET/computed tomography (CT) reports were collected and interpreted; 241 of these were iPET and 376 were ePET reports; 187 patients had both iPET and ePET performed. Nine expert clinical hemato-oncologists were asked to interpret the reports retrospectively. Three different hematologists separately evaluated each report, and results of interpretation were expressed as positive, negative or “intermediate.” Reviewers were blinded for all clinical data that could possibly affect their judgement, and were not allowed to communicate or comment on the reports with any of the nuclear medicine experts who initially examined and reported the scans.

The practical limitations of a retrospective analysis performed 4–8 years later on scans whose significance and prognostic implications were not entirely clear at the time of patient scanning (iPET), or for which at least two different interpretation keys were proposed several years later (ePET), are clearly evident. Moreover, in this study, most of the iPET scans (142/241: 59%) were reclassified or reinterpreted by clinicians as “intermediate,” and the boundaries with a positive at one side and a negative scan at the other side were totally arbitrary. However, despite the above considerations and limitations, this analysis is still very interesting, since it provides important information on both the use and the clinical context of a frequently used imaging technique in daily clinical practice.

The awareness of the significance of the prognostic value of interim PET in DLBCL first started with the French experience in a retrospective cohort of patients with DLBCL, treated with a CHOP or CHOP-like regimen, with or without rituximab [2]. In this study, after completion of induction, 83% of PET-negative patients achieved a complete remission compared with only 58% of PET-positive patients. The interpretation key for iPET included both the number of sites involved and the intensity of the fluorodeoxyglucose (FDG) uptake in the residual foci. However, this was not that easily reproducible, mainly because of the arbitrary grading of the residual FDG uptake. Apart from difficulties with the interpretation criteria, the prognostic role of interim PET in DLBCL was later confirmed in a large meta-analysis: the sensitivity and specificity in predicting treatment outcome were 0.78 (95% confidence interval [CI], 0.64–0.87) and 0.87 (95% CI, 0.75–0.93), respectively [3]. In 2009 an international workshop of hemato-oncologists and nuclear medicine experts held in Deauville, France, simple and reproducible criteria for iPET interpretation by visual assessment were proposed for Hodgkin lymphoma (HL) and DLBCL [4]. Here the residual FDG uptakes assessed in the foci of persisting disease were compared to the FDG uptake in the so-called “reference” organs (mediastinal blood pool and liver), and scored along a five-point scale. Thereafter, validation studies were proposed, which eventually confirmed the feasibility and reproducibility of the so-called “Deauville five-point scale” [5,6]. In the meantime, in order to increase its specificity and positive predictive value, semiquantitative assessment of PET scans employing the standardized uptake value (SUV) was increasingly employed [7,8]. According to experts, semiquantitative results could be expressed as ΔSUV_MAX, representing the difference between maximum SUV (SUV_MAX) of the baseline and interim PET scans. However, the most important limitation of this method of analysis for these PET results in a multicentric study was the variability of SUV measurement in the absence of a basic and more reliable program for single scanner quality assurance and quality control, to ensure inter-scanner calibration [9,10]. Very recently the same concepts were stressed during a closed workshop on lymphoma staging and restaging, held in Lugano during the 12th International Congress on Malignant Lymphoma [11].

The ePET has been one of the first utilized applications of PET for clinical management of lymphoma. In a recent meta-analysis in a limited number of patients with aggressive lymphoma, the reported ranges for the sensitivity and specificity of ¹⁸F-FDG PET in predicting disease relapse at the end of therapy were 0.33–0.77 and 0.82–1.00, respectively. For a single residual mass evaluation, sensitivity ranged from 0.33 to 0.87 and specificity ranged from 0.75 to 1.00, respectively [12]. In 2007 a panel of experts first proposed incorporating ePET in the diagnostic work-up for treatment response evaluation for patients affected by lymphoma [13], and interpretation criteria were also proposed in a companion paper [14], combining anatomical (dimensional) and functional parameters. These criteria, however, were validated only in a small cohort of 54 patients [15]. Moreover, dimensional criteria for lymph node size assessment were not fully reproducible, showing the least accuracy for diameters between 15 and 20 mm [16]. In the above-cited closed workshop on lymphoma staging and restaging during 12th Lugano meeting, the Deauville five-point scale, originally recommended for iPET, was also proposed as an interpretation key for ePET. It was also suggested that the threshold value for a negative versus a positive ePET should be set between scores 3 and 4, with the warning that patients with a score of 3, probably representing a complete metabolic response (CMR), may have different outcomes from those in patients scored as 1–2, depending on the clinical context and treatment regimen used [11].

All the above considerations should be kept in mind when interpreting the results of the study reported here by Mylam and co-workers [1], in which the only predefined criterion for iPET and ePET interpretation was the agreement among the three reviewers. On the basis of this simple interpretation key, each PET/CT report was “centrally” labeled as either positive or negative only if all three interpreters independently agreed, while in the remaining cases all other conclusions were considered “indeterminate.” One could argue that the only reason for a central review by three reviewers using such a “liberal” criterion for PET interpretation was merely to reproduce what is currently thought and performed in clinical practice by the treating hemato-oncologists. The fact that up to 59% of the iPETs were adjudged as “intermediate” basically reflects the concept that iPET has only a limited value in the clinical management of patients with DLBCL. Moreover, the reported 2-year progression-free survival (PFS) rates for positive, indeterminate and negative iPET of 52%, 85% and 90% respectively, with a very similar treatment outcome for patients with a negative and an indeterminate report, indicate a poor specificity and a very low positive predictive value for the set of interpretation criteria used. The same was true for ePET assessment: the 2-year PFS rates of patients with positive, indeterminate and negative ePET reports were 36%, 86% and 95%, respectively, with no difference in PFS between the “indeterminate” and negative subgroups. Once again, this suggests a low specificity of ePET, especially in cases with a single residual FDG-avid mass. In summary, while the five-point score seems to be widely accepted and also feasible for iPET in HL [17,18], the study by Mylam et al. implies that interpretation criteria for both iPET and ePET in DLBCL are still unsettled.

Potential conflict of interest

Disclosure forms provided by the authors are available with the full text of this article at www.informahealthcare.com/lal.