Annals of Nuclear Medicine

, Volume 25, Issue 10, pp 701–716

Impact of FDG-PET/CT in the management of lymphoma

Authors

    • Department of Clinical Radiology, Graduate School of Medical SciencesKyushu University
  • Koichiro Abe
    • Department of Clinical Radiology, Graduate School of Medical SciencesKyushu University
  • Takuro Isoda
    • Department of Clinical Radiology, Graduate School of Medical SciencesKyushu University
  • Yasuhiro Maruoka
    • Department of Clinical Radiology, Graduate School of Medical SciencesKyushu University
  • Masayuki Sasaki
    • Department of Health Sciences, Graduate School of Medical SciencesKyushu University
  • Hiroshi Honda
    • Department of Clinical Radiology, Graduate School of Medical SciencesKyushu University
Invited review article

DOI: 10.1007/s12149-011-0549-0

Cite this article as:
Baba, S., Abe, K., Isoda, T. et al. Ann Nucl Med (2011) 25: 701. doi:10.1007/s12149-011-0549-0

Abstract

Since the introduction of 67Gallium-citrate 30 years ago, nuclear medicine has played an important role in the evaluation of malignant lymphoma. During that time, several radiotracers were evaluated as potential alternatives for the diagnosis of lymphoma, but the introduction of 18F-fluorodeoxyglucose PET (FDG-PET) marked a major turning point. FDG-PET took over most of the role of gallium, and is now an essential tool in the diagnosis of lymphoma. FDG-PET is increasingly being used for assessment of the tumor staging prior to treatment, for evaluating the response to treatment, and for monitoring the early reactions to therapy to predict the final outcome. FDG-PET has been shown to have more accurate diagnostic capability than conventional CT and MRI for distinguishing the tumor necrosis and residual masses frequently seen after therapy in lymphoma patients without any clinical and biochemical manifestation. Malignant lymphoma is the first disease for which FDG-PET was adopted as a tool for response assessment in the international standard criteria. However, lymphoma does not always display a clear high uptake, and there are some pitfalls in assessing the response to therapy. This review will highlight the most important applications of FDG-PET in lymphoma, focusing on the advantages and pitfalls of this imaging, and past and ongoing efforts to standardize the use of FDG-PET, particularly in response to assessment and therapy monitoring.

Keywords

Malignant lymphomaResponse assessment of lymphomaFDG-PET/CT in lymphomaPretreatment staging of lymphomaTherapy monitoring of lymphoma

Introduction

The malignant lymphomas, Hodgkin’s lymphoma (HL) and non-Hodgkin’s lymphoma (NHL), are the tenth most frequently occurring type of cancer in Japan and increasing in the prevalence. Once the diagnosis of HL or NHL has been established by biopsy, determination of the disease extent (staging) is important for appropriate treatment planning and for determining prognosis. For nearly three decades, conventional imaging modalities, mainly using CT, have been used as the primary imaging tools to assess lymphoma. However, some problems have also been reported in the CT-based evaluation of response to therapy [17]. This is mainly due to the fact that CT cannot distinguish viable remaining tumor from benign residual mass mainly composed of necrosis or fibrosis. These residual masses (RMs) are seen after therapy in one-third of patients with NHL and two-thirds of those with HL [3, 8, 9]. Introduction of the first molecular imaging agents (i.e., 67Ga-citrate) has improved the accuracy of the response evaluation of patients with lymphoma [47]. However, the positive and negative predictive values of 67Ga are not sufficiently high. With its lower spatial resolution, low specificity and low sensitivity (especially in low grade lymphoma), 67Ga sometimes cannot distinguish abdominal uptake from physiological uptake of the bowel. Because of these drawbacks, 67Ga has been replaced by positron emission tomography (PET) using 18F-fluorodeoxyglucose (FDG), which estimates glucose metabolism. The sensitivities of FDG-PET for the detection of nodal and extra-nodal lesions of lymphoma are 92–100 and 74–78%, respectively, much higher than the corresponding values for 67Ga (44–48 and 42–52%, respectively) (Fig. 1) [10, 11]. FDG-PET is now becoming an essential tool to evaluate the treatment response in lymphoma. This review outlines the clinical application of FDG-PET, which can be broadly classified as primary staging before therapy, assessment of treatment response, early monitoring of therapy, and surveillance after therapy.
https://static-content.springer.com/image/art%3A10.1007%2Fs12149-011-0549-0/MediaObjects/12149_2011_549_Fig1_HTML.jpg
Fig. 1

Comparison of a 67Ga-scintigram (a) and 18F-FDG-PET scan (b) of the same patient performed within 1 week. A huge abdominal mass was visualized by both modalities, but it was very hard or impossible to pick up the left supraclavicular lesion (arrowhead) and small peritoneal lesion (arrow) by the 68Ga-scintigram, although these lesions were clearly visualized by FDG-PET. The pretreatment staging was advanced based on the PET findings

Detection of lymphoma

Diagnosis of lymphoma is basically made by biopsy. However, tomographic images such as CT and MRI have recently become more popular, and the likelihood of making an initial differential diagnosis by these modalities has increased. Common radiographic features of lymphoma include the following: homogeneous lymph node lesions without necrosis, continuous or scattered whole-body distribution of nodal lesions, splenomegaly and intra-spleen lesions, non-destructive expansion of extranodal lesions, invasive progression without occlusion of the adjacent gastrointestinal tract or vessels due to the soft nature of the tumor, and renal cortical lesions that are rarely seen in other malignancies [12]. In particular, the formation of multiple lymph node lesions in distant sites is one of the characteristics differentiating this disease from lymph node metastasis of common cancer, and thus the fact that FDG-PET allows whole-body scanning is a strong point in its favor.

Malignant lymphoma generally has a higher cell density than other cancers, and thus many lymphomas show high FDG uptake. The degree of uptake is largely dependent on the histology of lymphoma (with the aggressive type usually exhibiting higher uptake) (Fig. 2). In a study that classified 766 cases of lymphoma as FDG-avid or FDG-non-avid lymphoma, the sensitivity of FDG-PET was 100% for Hodgkin’s disease (233/233), Burkitt’s lymphoma (18/18), mantle cell lymphoma (14/14), nodal marginal zone lymphoma (MZL) (8/8), and lymphoblastic lymphoma (6/6), 97% for DLBCL (216/222) and 95% for follicular lymphoma (133/140) [13]. In the study of Paes et al. [14], the lymphoma subtypes were classified into four categories according to their FDG uptake. In another report, FDG-PET exhibited lower sensitivity for detecting extranodal involvement of lymphoma disease than CT in small lymphocytic lymphoma (SLL), lymphoid tissue accompanying mucosa (MALT), and marginal zone lymphoma (MZL) [15]. FDG-PET is likely not suitable for these lymphoma subtypes, and CT alone would be sufficient for their staging. A limited number of studies are available for peripheral T-cell lymphoma (PTCL) and related subtypes. In this category, systemic anaplastic large cell lymphoma (ALCL) usually displays avid FDG. The FDG avidity of PTCL-U (unspecified) has varied widely among reports [9, 1618]. Compared with FDG-avid skin lesions of advanced-stage cutaneous T-cell lymphoma (CTCL), the skin lesions of early-stage CTCL or ALCL are difficult to visualize by FDG-PET [9, 17, 18]. Uptake of FDG of various lymphoma subtypes is summarized in Table 1 [9, 1318].
https://static-content.springer.com/image/art%3A10.1007%2Fs12149-011-0549-0/MediaObjects/12149_2011_549_Fig2_HTML.jpg
Fig. 2

Typical findings of lymphomas of various pathology subtypes. Each patient had bilateral wide-spread nodal lesions, showing different degrees of uptake. The uptake level of lymphoma increased according to its pathological aggressiveness. a Follicular lymphoma; b diffuse large B-cell lymphoma; c lymphoblastic lymphoma

Table 1

Lymphoma subtype and FDG accumulation

Lymphoma

Subtype

FDG accumulation

Hodgkin lymphoma

Nodular sclerosis (grades 1, 2)

High

Mixed cellularity

Moderate to high

Lymphocyte depletion

Moderate to high

Classical Hodgkin lymphoma, lymphocyte rich

Low

B-cell lymphoma

Diffuse large B-cell lymphoma

High

Burkitt lymphoma

High

Follicular lymphoma (grade 3)

Moderate to high

Extranodal marginal zone B-cell lymphoma of MALT

None to high

Mantle cell lymphoma

Low to moderate

Follicular lymphoma (grades 1, 2)

Low to moderate

Lymphoplasmacytic lymphoma

Low

Nodal marginal zone B-cell lymphoma

None to high

B-cell chronic lymphocytic leukemia/small lymphocytic lymphoma

None to high

T-cell and NK-cell lymphoma

Extranodal NK/T-cell lymphoma, nasal-type

High

ALCL, primary systemic type

High

CTCL advanced

High

PTCL (unspecified)

Low to high

ATLL

Moderate

Sezary syndrome

Low

Mycosis fungoides

Low

CTCL early

Low

This table is originated from Ref. [15] with some addition

MALT mucosa-associated lymphoid tissue, NK natural killer, ALCL anaplastic large cell lymphoma, PTCL peripheral T-cell lymphoma, ATLL adult T-cell lymphoma/leukemia, CTCL cutaneous T-cell lymphoma

Staging of lymphoma

Using FDG-PET, the presence of lymphoma disease can be evaluated with greater certainty. The detectability of malignant lymphoma lesions by FDG-PET has been reported to be 70–100%, or about 15% higher than that by conventional CT, since the latter modality is dependent on the tumor size and shape. According to a meta-analysis, the median sensitivity of FDG-PET is 90% per patient unit and 97% per lesion unit [19]. Especially in the case of Hodgkin’s lymphoma, FDG-PET has been reported to have a higher sensitivity for the detection of extranodal lesions and bone marrow lesions compared to conventional methods, and these lesions have been upstaged in 15–25% of cases.

FDG-PET has also been shown to be quite sensitive in detecting nodal or extra-nodal lesions in patients with suspected lymphoma [2031]. Compared with CT, FDG-PET detects more lymphoma lesions, and particularly extra-nodal lesions (e.g., liver, spleen, bone marrow and muscle), in patients with DLBCL, follicular lymphoma (FL) and HL [2031]. Pretreatment assessment using FDG-PET changes tumor staging in about 5–15% of patients, and changes treatment strategy in 10–20% of patients compared with the conventional CT-based staging method (Table 2) [22, 23, 2527, 29, 30, 3235].
Table 2

FDG-PET for staging lymphoma

Authors (ref)

No. of patients

Upstage (%)

Downstage (%)

Change in therapy (%)

HL

 Buchman et al. [22]

25

8

0

8

 Wirth et al. [32]

31

14

0

18

 Pelosi et al. [33]

35

11.4

 

9

 Naumann et al. [29]

88

14.7

8

18

NHL

 Bangerter et al. [23]

44

12

2

14

 Partridge et al. [26]

44

40.9

<10

25

 Jerusalem et al. [25]

33

1

1

1

 Weiharauch et al. [27]

22

18

0

5

 Munker et al. [34]

73

29

3

<1

 Hutchings et al. [30]

99

17

5

7

 Rigacci et al. [35]

186

14

1

7

PET positron emission tomography, HL Hodgkin’s lymphoma, DLBCL diffuse large B-cell lymphoma, NHL non-Hodgkin’s lymphoma, PSF progression-free survival

PET/CT

Imaging by PET is vastly improved by combining it with CT technology. FDG-PET and CT provide functional and anatomic information, respectively. Integration of both modalities can compensate for the disadvantages of each, and the combination method outperforms either FDG-PET alone or CT alone for the staging of malignant lymphoma. Metabolic information provided by PET helps in detecting the evidence of disease which is undetectable by CT (e.g., focal uptake of spleen, non-distractive bone lesion). On the other hand, CT compensates for the poor spatial resolution of PET, thereby reducing the rate of false positives (e.g., due to brown fat), and helps in picking up PET-negative small lesions [36]. In a study of 50 cases of NHL that was designed to test the diagnostic performance of CT, PET and PET/CT, CT was found to be inferior to PET in detecting extra-nodal disease and head and neck lesions. PET alone and CT alone were inferior to PET/CT in detecting paraaortic and parailiac lesions [37]. In other studies, FDG-PET/CT correctly downstaged a number of patients compared with both CT and FDG-PET. FDG-PET/CT is associated with fewer false positives than FDG-PET alone, especially in the deep nodal regions of the abdomen and the mediastinum [30, 36]. Thus, PET/CT is more accurate in the evaluation of lymphoma than PET alone. Therefore, it should be remembered that many of the previously reported large-scale studies on the sensitivity and specificity of PET for detecting lymphoma were performed in the age before PET/CT introduction, and thus their results might be substantially different from the current PET/CT data.

There is a question as to whether or not PET/CT with non-contrast CT is adequate or if there is a need for contrast-enhanced CT (CE-CT). Use of the latest PET/CT scanner with i.v. contrast will result in equally good or better information compared to using PET and CE-CT at the same time. A recent study supports the idea that FDG-PET/CT using the CT portion with i.v. contrast is preferable to PET/CT without i.v. contrast [38]. Some studies have compared the diagnostic performance of PET/CT with a typical reduction in the CT radiation dose, and those with full-dose CT with i.v. contrast and concluded that the latter is superior for detecting extra-nodal disease. The authors of these studies considered that the improvement was largely attributable to the i.v. contrast rather than the use of full-dose CT. However, other reports suggest that low-dose PET/CT is adequate as an alternative to CE-CT even if it shows some limitations for assessing the liver and spleen. This means that most of the lesions of the liver and spleen that are undetectable by non-contrast CT are detectable by PET, and there are few lesions that are detectable only by CE-CT and not by PET (Fig. 3) [31, 39, 40]. It still remains a subject of controversy whether low-dose non-contrast PET/CT should be used or high-dose contrast PET/CT. This issue will have to be investigated in a large multi-center trial in the future.
https://static-content.springer.com/image/art%3A10.1007%2Fs12149-011-0549-0/MediaObjects/12149_2011_549_Fig3_HTML.jpg
Fig. 3

A case of splenic infiltration of DLBCL visualized by FDG-PET/CT (a). The same infiltration was difficult to detect by diagnostic contrast-enhanced CT (b)

Pitfalls in the diagnosis of lymphoma patient with FDG-PET

There are several pitfalls in the pretreatment staging of lymphoma using FDG-PET. The clinician must be aware that diffuse high FDG uptake of the bone marrow can occur due to the reactive change of chemotherapy, and in such cases the FDG uptake will not necessarily represent diffuse infiltration of the lymphoma [41]. Colony-stimulating factors such as G-CSF, which is often used in blood disorders, accelerate the metabolism of bone marrow, and hyperactive marrow shows high FDG uptakes mimicking disease. In cases with prior radiotherapy or an underlying disorder of the bone marrow such as myelofibrosis, the FDG uptake pattern of the activated marrow will be more complex and hard to distinguish from lymphoma disease. Degenerative changes may also be mistaken for bone metastases due to the focal high FDG uptake. On the other hand, minimal bone marrow disease may be missed by FDG-PET, and thus FDG-PET should not be considered a replacement for bone marrow biopsy [16, 42]. Because of these pitfalls, treatment planning should not be changed solely based on FDG-PET findings. Before a change in treatment planning, the PET findings should be reconfirmed by another imaging modality and/or biopsy [41, 43]. False-positive findings of FDG-PET may occur for multiple reasons, and therefore PET detects more lesions than CT. Conversely, mass lesions identified by CT may present as negative by PET. It is clear that PET alone cannot replace CT for the purposes of pretreatment staging of lymphoma.

FDG-PET for treatment response assessment

Response to initiated treatment serves as a surrogate for other clinically significant parameters, such as progression-free survival (PFS) and overall survival (OS). Response is also an important guide in making a decision whether to continue or change the ongoing therapy. Until recently, response evaluation in lymphomas was performed according to the International Workshop Criteria in NHL and according to the Cotswold Criteria in HLL [8, 44]. However, there was a problem in assessing the RMs. RMs are often seen by CT or MRI after the completion of lymphoma therapy, and this remains a problem because it is hard to know whether these tumors are still viable or not. According to the meta-analysis of the accuracy of FDG-PET for the treatment response, PET can diagnose HL and NHL with a specificity of 90 and 100%, respectively. So FDG-PET is useful to distinguish remaining viable tumor from inactive scar tissue, and is helpful in determining the need for additional therapy (Fig. 4). Therefore, when we see remaining tumor, negative FDG-PET allows elective follow-up and positive PET suggests the need for additional aggressive treatment.
https://static-content.springer.com/image/art%3A10.1007%2Fs12149-011-0549-0/MediaObjects/12149_2011_549_Fig4_HTML.jpg
Fig. 4

Example of a residual mass after the completion of full-dose chemotherapy for a DLBCL patient. Compared with the pretreatment scan (a), the soft tissue mass had shrunk, but it was still visible by contrast-enhanced CT in the paraaortic region, with no uptake by FDG-PT/CT (b) regarded to achieved CR

According to the definition of CMS (Centers for Medicare and Medical Services) in the United States, re-staging after therapy is different from response monitoring of therapy, and is usually done after completion of 6–8 cycles of full-dose chemotherapy or to assess the suspected recurrence. The latter type of assessment is usually performed in the early period of therapy after 2–3 cycles of chemotherapy [36, 45]. Treatment response assessment is the most popular application of FDG-PET in HL and NHL patients, and is compatible with recently introduced response criteria [1, 9].

FDG-PET boasts a high NPV of greater than 80% in virtually all the reported studies [1, 9, 24, 41, 43, 4559]. Based on the findings of RMs with negative FDG-PET in FDG-avid-type lymphoma patients, recurrence can be ruled out with high confidence. The fact that 10–20% of the findings by PET scan are false-negatives (depending on the effectiveness of the treatment and histological sub-types of lymphoma) suggests that PET scanning often overlooks subtle tumors that may cause future recurrence. The positive predictive value (PPV) of FDG-PET is lower than that of NPV, and varies among reports.

When using a visual assessment, PPV of PET is 70–80% in aggressive NHL, and 60–70% in patients with HL [1, 9, 24, 41, 43, 4559]. The relatively low PPV value of HL compared with aggressive NHL is probably due to the false-positive interpretation of inflammatory change caused by the radiotherapy previously used in a large number of the HL patients, and the hyperactive thymus often seen in younger patients that make up much of the population with HL [54]. Although PPV of FDG in aggressive NHL and HL are middle-range of value (80 and 50%, respectively), these values are substantially higher than the corresponding values for CT (about 40 and 20%, respectively), suggesting that FDG-PET is more accurate than CT in the response assessment of lymphoma [1, 9, 24, 41, 43, 4559]. The low accuracy of CT is mainly due to the fact that CT cannot distinguish residual active tumor from inactive fibrosis or necrosis.

In HL, about two-thirds of patients have residual mass (RM) on CT after the completion of therapy despite the absence of clinical or other biochemical signs of relapse. More than two-thirds of these patients are negative on PET, and relapse occurs in less than 10% of these patients [24, 4857, 59]. Thus, these patients can be followed up without problems. In aggressive NHL (e.g., DLBCL), one-third of patients have such RMs and again, more than two-thirds of these patients are negative on PET. Relapse occurs in 15–20% of these patients [24, 4857, 59]. Both in aggressive NHL and HL, about one-third of PET-positive patients are at increased risk of recurrence (60–70%). On the other hand, in 30–40% of patients with positive PET for RMs, the disease does not progress and recur. Histopathological confirmation is important before initiating a salvage therapy in cases with positive PET findings [1, 9, 24, 41, 43, 4559].

Most of the false-positive FDG-PET findings in RMs are due to post-therapy inflammation. Other false-positive findings caused by inflammatory processes are the result of infection or chronic inflammatory conditions such as sarcoidosis, and brown fat or activated rebound thymic hyperplasia [1, 9, 41].

This usefulness of FDG-PET has been widely accepted. The International Harmonization Project (IHP) has developed new recommendations for response criteria for aggressive malignant lymphomas, incorporating FDG-PET findings into the definitions of end-of-treatment response in FDG-avid lymphomas in 2007 [1, 9]. It has been proposed that, for FDG-avid-tissue types, the therapeutic efficacy will be determined by the combination of changes in the PET findings and changes in the size. Particularly noteworthy in regard to these new criteria is the elimination of the Cru category (i.e., the cases in which the tumor remains on the image but does not change its size over 3 months without treatment). Most of the cases classified as CRu will now be designated as CR if the RM is PET negative or PR if it is PET positive.

Elimination of CRu results in much better differentiation in outcome between CR and PR patients. A retrospective study using 54 aggressive NHL patients showed that PET not only doubled the number of CRs but also significantly enhanced the difference in PSF between CR and PR [60]. Other recent retrospective analyses confirm the superiority of the new response criteria in HL and aggressive NHL [60]. Other recent retrospective analyses confirm the superiority of the new response criteria in HL and aggressive NHL [61, 62]. These findings provided the rationale for incorporating PET into the revised criteria for FDG-avid lymphomas.

Another important issue in regard to this recommendation is the active incorporation of visual assessment. A residual mass with a GTV (gross tumor volume) greater than 2 cm should be diagnosed as PET positive when its uptake is higher than its mediastinal blood pool level by visual assessment. RMs with a size of 1.1–1.9 cm are regarded as positive when their uptake is higher than the surrounding tissue. In other words, IHP consensus concludes that visual assessment is optimal for the evaluation of treatment response and quantitative or semi-quantitative methods (such as SUV) are not necessary for this purpose.

However, according to a recent study with 50 lymphoma patients that included cases of aggressive NHL (n = 24) and HL (n = 26), a semi-quantitative approach was superior to a visual-based approach for the precise evaluation of RM viability [54]. A comparison of the visual-based and semi-quantitative evaluation is also described in the following paragraphs.

By consensus report, FDG-PET is strongly recommended for patients with FDG-avid and curable-type lymphoma (the primary end-point is complete remission), such as diffuse large B-cell lymphoma and Hodgkin’s lymphoma. On the other hand, FDG-PET is obviously not recommended for lymphoma without detectable FDG accumulation. Even with FDG-avid-type lymphoma, if the primary end-point is not complete remission, such as in cases of incurable aggressive lymphoma or indolent lymphoma, PET is usually not recommended.

A recent study raised some concern in regard to the use of FDG-PET for posttreatment evaluation to direct further treatment strategy. This randomized study divided its 160 HL patients, all of whom had RMs after ABVD therapy, into a group receiving additional radiotherapy and a group receiving no further treatment. There was a significant difference in the percent relapse within 18 months between the radiotherapy group (2.5%) and no therapy group (14%) [63]. Another study yielded results in the other direction, suggesting that radiotherapy could be safely omitted with a 94% negative predictive value in advanced-stage HL patients irrespective of whether they had RMs more than 2.5 cm, as long as the results of posttreatment PET were negative [64].

The new response criteria are not yet supported by a sufficient amount of clinical data. A long-term follow-up by multiple institutes is still needed, and the modification of treatment strategy based on FDG-PET results should be regarded as an experimental approach at this time [65].

FDG-PET for therapy monitoring and response-adapted therapy (Interim PET)

Tumor response to induction treatment is a surrogate for other important clinical parameters, including PFS and OS. An early reliable prediction of response to therapy may improve the overall outcome of treatment by separating high-risk poor prognosis patients from those with good prognosis. For the high-risk group, a more intensive regimen can be started at an earlier point that may improve the possibility of complete remission. For the good prognosis group, harmful side effects by unnecessary treatment might be reduced with a less intensive and less toxic regimen.

When determining the therapeutic effect using morphological imaging modalities such as CT, reduction in the size of the tumor is the most important factor [66, 67]. However, size reduction of the tumor is not necessarily an accurate predictor of outcome. In HL, treatment-sensitive tumor cells make up a very small fraction of the tumor volume, which is usually surrounded by a large number of reactive infiltrating cells [3]. The process of shrinking the tumor will depend on the lymphoma cell type and the condition of the patient’s immune system. More importantly, a period of time is required until morphological changes can be observed and a final decision can be made by CT. At that time point, modification of therapy might be less effective.

On the other hand, changes in tumor metabolism, such as glucose metabolism, take place relatively early during the induction of therapy. Several studies have shown that these early metabolic changes are a good predictor of final treatment response both in aggressive NHL and HL (Table 3) [6878]. If the sensitivity to chemotherapy or immunochemotherapy can be evaluated at an early point of treatment, a change in treatment strategy can also be made at an early time point for those patients who are expected to show an insufficient response.
Table 3

Interim PET studies

Authors (ref)

No. of patients

Cycles of therapy

PET negative

PFS (%)

PET positive

PFS (%)

HL

 Kostakoglu et al. [68]

23

1

74

100

26

12.5

 Hutchings et al. [69]

85

2–3

72

94

13

38

 Hutchings et al. [70]

77

2

79

95

21

31

 Zinzani et al. [71]

40

2

80

97

20

12

 Gallamini et al. [72]

260

2

81

95

19

14

 Markova et al. [73]

50

4

72

100

28

86

NHL

 Kostakoglu et al. [68]

24

1

58

100

42

 

 Jerusalem et al. [74]

28

2–3

82

100

18

30

 Spaepen et al. [75]

47

3–4

47

84

53

0

 Haioun et al. [76]

90

2

60

82

40

43

 Mikhaeel et al. [77]

121

2–3

41.7

93

43

30

 Zinzani et al. [71]

91

 

61.5

89

38.5

17

 Cashen et al. [78]

50

2–3

30

85

30

75

PET positron emission tomography, HL Hodgkin’s lymphoma, DLBCL diffuse large B-cell lymphoma, NHL non-Hodgkin’s lymphoma, PSF progression-free survival

Most of the conventional prognostic factors are based on pretreatment clinical data (such as the pretreatment stage, LDH, age, and performance status) and have been well established as predictors of survival in large-scale studies [44, 7981]. In contrast, interim FDG-PET can be used to evaluate the response to treatment of individual patients, allowing more personalized and risk-adapted treatment strategies.

Early or interim PET is usually performed after 1–4 cycles, typically 2 cycles, of induced chemotherapy or immunochemotherapy. Kostakoglu et al. [68, 82] suggested that early PET after only one cycle of chemotherapy has the same high predictive value.

There was a significant difference in 5-year PFS between the interim PET-positive group (39%) and PET-negative group (92%), by a retrospective analysis of 88 HL patients scanned after two or three cycles of (ABVD)-like chemotherapy [69]. These results were confirmed later in prospective studies by Hutchings et al. [70], Zinzani et al. [71] and Gallamini et al. [72], with the latter two studies focusing on advanced-HL patients alone. In all three studies, almost all (94–100%) of the patients who were interim PET positive had refractory disease or relapsed within 2 years, while all the early PET-negative patients entered a good remission and rarely relapsed (∼6%). The prognostic value of early FDG-PET has been reported to be superior to the prognostic value of conventional International prognosis scores [83].

The rationale for interim PET is based on the hypothesis that it is a precise predictor of the final treatment response of patients. There are many studies that support this hypothesis. The reported negative predictive value of interim PET is over 80% in aggressive NHL and 90% in HL [58, 69, 70, 74, 8289]. These reports suggest that an interim PET scan is at least as accurate as a PET scan after completion of therapy. Because of the high NPV value and significant difference of outcome between PET-positive and -negative patients, interim FDG-PET is increasingly being used in the sense of risk-adapted therapy [90] (Table 3).

The exception here is the relatively low (only 30%) positive predictive value (PPV) of interim PET in HL patients compared with relative high NPV (over 95%) [58, 69, 70, 74, 8289]. This means that a considerable number of RMs, which are supposed to be negative at the end of therapy, display positive PET scans after 2 cycles of chemotherapy. This indicates that new standards, different from the posttreatment assessment, are required for interim PET. Some reports indicate that PPV is dependent on the stage of lymphoma. According to studies by Hutchings et al. [69, 70], the tentative PPV of FDG-PET in cases of early-stage HL was only 25%, versus 94% for cases of advanced-stage HL. A high PPV of interim PET in advanced-stage HL was also recently reported by Gallamini et al. [83].

As mentioned above, the prognostic power of negative FDG is outstanding, and there is a consensus that FDG-PET provides an accurate prognosis in patients who have received 1–4 cycles of chemotherapy or immunochemotherapy. On the other hand, close attention should be paid when evaluating the patient prognosis using a positive interim FDG-PET scan by visual assessment, because the PPVs of such scans are limited. Biopsy is strongly recommended when the treatment strategy is scheduled to change according to positive interim PET findings.

It is noteworthy that visual assessment is fully incorporated in the interpretation of PET results in the new criteria. Semi-quantitative parameters such as SUV are widely used in clinical practice and are a convenient measure for approximating the state of the disease. However, the reproducibility of SUV is not guaranteed among different institutes, different PET cameras and even different serial scans from the same patient. This is partly due to the fact that PET/CT is still under continuous development, and the performance of the older and newer, state-of-the-art scanners differs substantially. It is not difficult to imagine that incorporating SUV (including thresholding) in the international criteria at this point is disallowed because of expected confusion. As long as the interpretation depends on visual assessment, the results will vary depending on the strictness of the radiologist interpretation. Some reports suggest the use of more lenient standards for the positivity for interim PET on RMs to improve the PPV without detriment to NPV. Such a lenient interpretation would likely include the idea that a minimum residual uptake (MRU) that is equal to, or slightly higher than that of the liver should be considered to be negative (Fig. 5) [69, 70, 83, 8589]. In a retrospective study of 85 HL patients after 2–3 cycles of first-line chemotherapy, 94% (68/72) of patients displayed completely negative results or a progression-free MRU for up to 3.3 years, whereas 39% (5/13) of patients showed PFS with apparently positive PET scans. More strict criteria, which did not allow any uptake on RMs, did not enhance the difference in PSF between the positive and negative group [69]. The interpretation of interim PET using the lenient criteria allowing minimal uptake is at least as accurate for predicting patient outcome as with post-therapy PET using the strict criteria [69]. Similar findings were also reported in patients with HL and aggressive NHL patients, suggesting this lenient criterion is more appropriate for assessing the interim PET [70, 83, 8689].
https://static-content.springer.com/image/art%3A10.1007%2Fs12149-011-0549-0/MediaObjects/12149_2011_549_Fig5_HTML.jpg
Fig. 5

Example of minimum residual uptake (MRU) after the therapy. A bulky FDG-avid mass (a) was reduced in size after 4 cycles of R-chop (b) and radiotherapy (c). Minimum FDG uptake was consistently visible after therapy. This patient remained relapse-free for 3 years

Attempts have been made to provide a more detailed classification for the visual assessment of FDG uptake level. A European group introduced a 5-point scale for the visual assessment of interim PET. The accumulation levels in the liver and mediastinum were defined as the reference background, and the uptake of RMs was classified into 5 categories relative to these standards. The 5-point scoring scale was thus as follows: 1 (no uptake), 2 (uptake slight or equal to that in the mediastinum), 3 (uptake greater than in the mediastinum but lower than in the liver), 4 (uptake moderately greater than in the liver at all sites), and 5 (markedly increased uptake at a particular sites and new sites of disease). Using these criteria, the GELA group has launched several clinical trials. This approach is more informative and reproducible than a visual-based positive–negative dichotomic approach [91].

There are some evidences that a semi-quantitative approach using SUV may be superior to a visual-based dichotomic approach. In a report by Lin et al. that enrolled 92 patients with DLBCL, a cut-off value of 65.7% for the rate of reduction in SUVmax between baseline and the end of the second cycle of a CHOP or CHOP-like regimen predicted event-free survival (EFS) with an accuracy of 76.1%. In contrast, the prediction accuracy was 62.5% when using visual criteria that allowed the MRU to take a negative value [86, 87]. In this study, a visual approach yielded a similar NPV (75%) as an SUV-based approach (74.1%); however, the PPV by the visual approach (50%) was far inferior to that by the SUV approach (81.3%). When using a cut-off value of 5.0 in SUVmax, the accuracy was the same as by the SUV percent reduction method [86, 87]. More recently, Casasnovas et al. reported the usefulness of the reduction rate of SUVmax for predicting the early prognosis value [91]. Semi-quantitative values such as SUVs are still useful and more informative compared with visual assessment, when used with the appropriate parameters and internal standardization such as the reduction rate of SUVmax. A semi-quantitative approach using SUV might be more important in the future.

When semi-quantitative values are used for the purpose of prognostic evaluation, standardization of PET is obviously very important. Only with this standardization, comparative study will be feasible [9]. SUV depends largely on multiple parameters such as spatial resolution, reconstruction parameters, etc. Large-scale clinical trials require multi-center studies, and thus common interpretation criteria are needed.

Standardization should be made both in data acquisition (including image processing) and image interpretation. This standardization of PET for lymphoma is a new phase of clinical application of PET and will be applied to other malignancies in the near future.

In the study of Meignan et al., there was a difference in the bio-physiological features between HL and aggressive NHL. The GELA group proposed two different sets of criteria for interim PET interpretation performed in HL and DLBCL, respectively. This was based on the assumption that the cellular architecture and physiopathology of the neoplastic tissue was different in these two types of lymphoma subsets. In HL, neoplastic Reed–Sternberg (RS) cells account for less than 1% of the overall cellularity of the neoplastic tissue, whereas in NHL, they contribute more than 90% of the total cell population. In HL, bystander, non-neoplastic lympho-mononuclear cells produce a cytokine network that ensures the immortalization of RS cells and works as an amplifier of the PET detection power. This non-neoplastic cellular compartment is switched-off very early by chemotherapy: this phenomenon is also known as “metabolic CR.” On the other hand, in DLBCL a progressive fraction of neoplastic cells are lysed by the chemotherapy, and the percentage of the cell destruction is predictive of the final response to the chemotherapy. For these reasons, a visual assessment seems preferable in HL, whereas a quantitative approach by SUVmax measurement seems more appropriate in DLBCL [92].

Ongoing risk-adapted trials

HL

A risk-adapted approach will be very effective when multiple therapeutic strategies can be selected according to patient status, and effect/side effect is trade-off relationship. HL is suitable for this concept. Early-stage HL patients are usually cured with standard ABVD [Adriamycin (doxorubicin), bleomycin, vinblastine, and dacarbazine] therapy. However, the life expectancy of those patients is reduced due to the side effects of chemotherapy even after remission of the HL disease. According to a report, more early-stage HL patients die from the late effects of therapy than from the original disease [93]. This means a substantial number of early-stage HL patients might be subject to over-treatment. When adequate therapeutic effects are expected, the treatment dose might be reduced without compromising the therapeutic outcome. This is the background for using interim FDG-PET to identify good-risk patients eligible for less intensive therapy. More specifically, reducing the total cycles of ABVD or omitting radiotherapy may be appropriate in these patients.

Multiple clinical trials using PET response-adaptive therapy are now ongoing in Europe (Table 4). The non-inferiority of reducing treatment intensity by omitting radiotherapy to interim PET-negative patients is currently being investigated by the UK National Cancer Research Institute (NCRI) Lymphoma Group RAPID trial for early-stage patients, and the German Hodgkin Study Group (GHSG) HD16 protocol and EORTC/GELA/IIL H10 protocol [9496].
Table 4

Ongoing risk-adapted clinical trial

Patients

ClinicalTrials.gov identifier

Study group

PET timing

PET result

Intervention

Study type

Early-stage HL

NCT00736320 (HD16) [94]

German Hodgkin Study Group

2 × ABVD

Negative

No radiotherapy

Phase III

NCT00433433 (H10 trial) [95]

EORTC/GELA/IIL

2 × ABVD

Negative

No radiotherapy

Phase III

NCT00282035 (RAPID trial) [96]

UK NCRI lymphoma group

3 × ABVD

Negative

No radiotherapy

Phase III

Advanced HL

NCT00795613 [98]

GITIL

2 × ABVD

Positive

BEACOPPesc

Phase II

NCT00678327 (RATHL) [99]

UK NCRI lymphoma group

2 × ABVD

Positive

Intensification BEACOPP

Phase IIIa

NCT00784537 (HD0801) [100]

IIL

2 × ABVD

Positive

Salvage regimen

Phase IIIa

NCT00515554 (HD18) [101]

German Hodgkin Study Group

2 × ABVD

Negative

4× vs. 8× BEACOPPesc

Phase III

DLBCL

NCT00530179 (PET CHOP) [104]

Alberta Cancer Board

2 × R-CHOP

Positive

Salvage with HD + ASCT

Phase II

NCT00144807 (LNH2007-3B) [105]

GELA

2 × R-CHOP

Positive

Salvage with HD + ASCT

Phase IIIa

NCT00324467 [106]

British Columbia Cancer Agency

4 × R-CHOP

Positive

4 cycles R-ICE

Phase II

Aggressive NHL

NCT00238368 [107]

Johns Hopkins

2–3 × R-CHOP

Positive

Salvage with HD + ASCT

Phase II

NCT00554164 (PETAL) [108]

University Hospital, Essen

2 × R-CHOP

Positive

(R-)CHOP vs. Burkitt regimen

Phase III

ABVD doxorubicin, bleomycin, vinblastine and dacarbazine, ASCT autologous stem cell transplantation, BEACOPPesc bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine and prednisone, DLBCL diffuse large B-cell lymphoma, EORTC European Organisation for the Research and Treatment of Cancer, FDG 2-18F-fluoro-2-deoxyglucose, GELA Groupe d’Etudes des Lymphomes de l’Adulte, GITIL Gruppo Italiano Terapie Innovative nei Linfomi, HD high-dose chemotherapy, HL Hodgkin lymphoma, IIL Intergruppo Italiano dei Linfomi, NCRI National Cancer Research Institute, NHL non-Hodgkin lymphoma, RAPID response-adapted therapy with early PET in Hodgkin’s disease, R-CHOP rituximab, cyclophosphamide, doxorubicin, vincristine and prednisone, R-ICE rituximab, ifosfamide, carboplatin and etoposide

aNo randomization regarding PET response-adapted therapy

Most (approximately 70%) patients are reported to be cured with ABVD with or without prolonged cycle or consolidation radiotherapy, which is the first-line therapy in most centers. The more intensive BEACOPPesc regimen (bleomycin, etoposide, Adriamycin, cyclophosphamide, vincristine, procarbazine, and prednisone) cures 85–90% of patients if given upfront, but the concerns regarding side effects are the reason why many centers hesitate to use this regimen as standard therapy [97]. Advanced-stage patients who have poor response to induced first-line therapy have a much worse prognosis. For such patients, an early switch to an escalated BEACOPP regimen may increase the chances of a good long-term prognosis (GHSD HD9 trial). Escalated BEACOPP is myelosuppressive and is associated with an increased risk of secondary malignancies. Therefore, the accurate selection of these high-risk patients is of a great significance. To lower the risk of treatment failure, avoid unnecessary toxicity and increase the chance of long-term survival, these non-responders should be identified as early as possible [98].

Clinical trials of PET response-adapted therapy for advanced-stage HL patients have also been launched. For patients who are still PET positive after two cycles of ABVD, early intensification with BEACOPPesc is applied in most of the trials (Italian GITIL trial and the European RATHL trial) [99, 100]. Autologous stem cell transplantation is also applied for those patients (ASCT; Italian IIL trial) [101]. Another ongoing BEACOPPesc-based study is focused on testing the non-inferiority of the reducing dose. In that trial, advanced-stage HL patients are randomized into an abbreviated treatment course if the post-two cycle interim PET is negative [102].

Based on recent studies, the PPV of interim FDG-PET may be lower in patients treated with the BEACOPPesc regimen than in patients treated with ABVD [103, 104]. This is partly due to the fact that the PPV of interim FDG-PET may be maintained at a high level among advanced-stage patients. Only one out of five interim FDG-PET-positive early-stage patients relapsed, and this patient was successfully salvaged [70]. Therefore, intensification, with or without BEACOPPesc, in early-stage HL patients with a positive interim FDG-PET may cause unnecessary toxicity in patients who would have been cured with standard therapy.

NHL

Several studies have reported a substantial difference in the progression-free survival between interim PET-positive patients (10–50%) and PET-negative patients (79–100%). A study with 121 high-grade NHL patients concluded that the response on FDG-PET after two or three cycles of R-CHOP is predictive of PSF and OS. The estimated 5-year PSF was 89% in the negative group, 59% for patients with minimal residual uptake (MRU) on FDG-PET and 16% in the PET-positive group [77]. According to another report on 90 prospectively studied patients with aggressive NHL, patients with a positive interim PET after 2 cycles of chemotherapy had a 2-year PSF rate of 43% and OS rate of 60%, versus corresponding values of 82 and 90% in a PET-negative group [76]. When compared with the prognostic power of IPI, interim PET after 2 cycles of chemotherapy was superior for predicting PSF and OS (the p values were <0.03 and <0.58, respectively) [75]. These results are somewhat inferior to those with HL. The lower NPV was likely attributable to the intrinsically worse prognosis of the disease. The lower positive predictive value may have been partly due to the higher risk of infection related to dense-dose chemotherapy because of the higher median age of patients with NHL compared to those with HL.

Aggressive NHL patients who respond poorly to first-line treatment or relapse soon have a very poor prognosis and generally require high-dose salvage therapy. Such patients could benefit from early recognition of the treatment failure. Similar to HL, a multi-center clinical trial on aggressive NHL has also been launched by a group in the United States and Europe [83].

If the information provided by the results of these clinical PET studies is proven to affect the final outcome of the patient, interim FDG-PET will play an increasingly prominent role in the clinical setting. Several studies have focused on whether interim PET-positive DLBCL patients would benefit from an early escalation to more intensive regimens or even high-dose therapy with ASCT [105109]. The problem here is that the R-CHOP is now the first-line and the only primary therapy for aggressive NHL, and fewer optional regimens are available than in HL. The interim report of SWOG S9704, which compared patients who received 8 cycles of CHOP and those who received 6 cycles of CHOP followed by ASCT, concludes that there was no significant difference in OS between the two groups, suggesting there might be less benefit of PET-based stratification of patients than in HL (ASCO2011). A recent report from Korea using 155 DLBCL patients concluded that the usefulness of interim PET is skeptical in predicting overall survival [110].

FDG-PET for post-therapy surveillance

A surveillance FDG-PET scan is performed for the purpose of early detection of recurrence in cases without any clinical, biochemical or radiologic evidence of recurrence. Application of FDG-PET in this context remains controversial, in that there is potential to increase costs without proven benefit because of false positives that exist to some extent in PET [111]. There have been relatively few studies investigating the clinical value of FDG-PET for the follow-up surveillance of lymphoma. In an early study by Dittmann et al., 21 HL patients were retrospectively studied and found that FDG-PET and CT were equally sensitive in detecting relapse before the occurrence of symptoms [55]. In a study that followed 36 HL patients using PET every 4–6 months for 2–3 years after the completion of therapy, FDG-PET detected 11 positive patients. In 5 of these 11 patients (1 who had a viable residual mass and 4 who relapsed during 5–24 months) treatment was deemed to have failed by FDG-PET before the appearance of clinical symptoms or signs, laboratory results or CT imaging suggested relapse. Confirmation of relapsed disease was obtained by biopsy in 4 patients and by CT findings and clear clinical symptoms in the remaining patient. However, in 6 patients, false-positive PET results incorrectly suggested possible relapse, necessitating an additional re-staging scan, although these findings were later confirmed to be negative in all cases [90].

Recently Zinzani et al. reported more large-scale surveillance study of lymphoma patients using FDG-PET. A total of 421 lymphoma patients (160 with HL, 183 with aggressive NHL, and 78 with indolent NHL) were followed up by FDG-PET at 6, 12, 18 and 24 months after completion of treatment and every 1 year after first CR. The percent detection rate of relapse was higher for FDG than by CT or by conventional clinical manifestations (HL 32, 23, 22%; aggressive NHL 31, 25, 22%; indolent NHL 60, 49, 38%, respectively). In 36 cases, additional biopsy was needed due to the inconclusive results of FDG-PET. Relapse was diagnosed correctly by PET in 24 cases (67%), but in 12 cases (33%) the results were false-positive, including 9 cases with reactive lymph nodes and 3 with sarcoid granulomas [112]. However, these studies did not provide information as to whether or not surveillance PET was cost-effective, and moreover whether it contributed to an improvement of the overall prognosis. Therefore, surveillance FDG-PET cannot be recommended for use beyond the clinical trial stage, at least until a more definitive conclusion is reached by a large prospective cohort study [90, 112].

FDG-PET has some ability to distinguish indolent lymphoma and aggressive lymphoma, indicating that FDG-PET can be used to confirm clinically suspected cell transformation of indolent-type lymphoma into more aggressive subtypes (large cell transformation). FDG-PET not only supports a clinical diagnosis of transformation, but also provides an optimal site for biopsy, which usually has distinguishable high uptake. However, FDG-PET cannot replace biopsy because there is considerably overlap between the indolent and aggressive subtypes [113, 114]. FDG-PET is still useful in such cases where biopsy is technically difficult or is relatively invasive. From the report by Schoder et al. [114] this SUV-based approach is useful when detected SUVmax falls in the range of typical uptake of aggressive and indolent lymphoma (<6, and >13, respectively). Here we have to pay attention that in only half of the patients transformation can be evaluated correctly by typical uptake profile, less than half the cases are difficult to evaluate because of overlapping intermediate range of uptake [114].

Conclusion

FDG-PET is currently the most accurate tool for the assessment of treatment response and prognosis in lymphoma. A negative PET excludes residual disease with reasonable high certainty. In contrast, positive PET findings may have to be confirmed by biopsy before treatment regimen is changed. Interim PET is increasingly used for response-adapted therapy modifications. There is considerable hope that these strategies may improve patient outcome. However, interim PET should be considered investigational and should not be used for patient management outside the study protocol.

Copyright information

© The Japanese Society of Nuclear Medicine 2011