Background

Multiple myeloma (MM) is a malignant plasma cell disorder that is currently incurable in most cases. The introduction of novel therapeutic agents significantly improved the progression-free and overall survival, but the vast majority of patients will eventually experience disease progression [1, 2]. The multifaceted presentation and progressive nature of MM require complex assessment including detailed radiological examination. Whole-body computed tomography (CT), magnetic resonance imaging (MRI), and positron emission tomography—computed tomography (PET/CT) are the methods of choice, depending on the clinical setting and their availability [3].

The 18F-fluorodesoxyglucose (18F-FDG) PET/CT is particularly useful for differentiating between active and smoldering myeloma, assessing residual disease after therapy, or detecting extramedullary involvement [4, 5]. Disease activity assessed by PET/CT at diagnosis, after induction therapy, and after high-dose therapy is a useful predictor of progression-free survival (PFS) and overall survival (OS), as shown by multiple prospective and retrospective studies (Table 1). Moreover, comparative pre- and post-therapeutic PET/CT for evaluation of therapy response yields important prognostic information, too [6, 7]. The IMWG consensus statement on radiologic diagnostics in MM recommends the use of PET/CT at diagnosis when available, especially in studies, followed by PET/CT re-evaluation after 1st line treatment for assessment of residual disease [3].

Table 1 Overview of studies evaluating prognostic role of variable PET-CT parameters in multiple myeloma
Full size table

Most studies on the prognostic role of PET/CT in MM measured metabolic disease activity using standardized uptake values (SUV) and evaluated the number of FDG-avid focal intramedullary lesions (FL) as well as the presence of extramedullary disease (EMD) [4, 8,9,10,11,12,13,14,15,16]. Prospective studies during first line therapy identified cut-off values for PET/CT parameters with minor variability [8, 11, 12, 15]. These prognostically significant thresholds were reproduced in multiple smaller retrospective studies (Table 1). A maximal SUV (SUVmax) above 3.9–4.2, more than 3 FLs, and the presence of PET-positive EMD were repeatedly demonstrated to be negative predictors for OS and/or PFS [8,9,10, 14, 15, 17].

The role of PET/CT in relapsed or refractory MM is less clear, particularly in the post-transplantation setting [4]. The parameters and thresholds used in relapse studies are heterogeneous, which hampers the evaluation of the prognostic role of PET/CT. The prognostic thresholds proposed for SUVmax and number of focal lesions in patients with relapsed MM are 3–4 times higher and more variable than those identified in the first-line setting [18, 19] (Table 1).

In our retrospective single center study, we analyzed patients with MM who underwent at least one HDT with autologous or allogeneic SCT and experienced disease relapse. We evaluated the prognostic role of 18F-FDG-PET-CT for OS and PFS, focusing on prognostic parameters that were mostly identified in larger prospective studies looking at baseline parameters and/or response to therapy.

Methods

We retrospectively evaluated patients with MM who underwent 18F-FDG-PET/CT in our center between 2012 and 2019. In this population, 36 pre-treated patients, who had received at least one autologous or allogeneic SCT and experienced disease relapse at the time of PET/CT investigation, were identified and included in the study. Relapse was defined according to IMWG-criteria [20]. Patients with relapse who had already started new therapy prior to the PET/CT were excluded. Clinical findings and laboratory results were integrated into our analysis (Table 2).

Table 2 Overview of patient’s data
Full size table

PET-CT Scans

All scans were performed as a non-contrast low-dose PET/CT (mCT, 128 slices, Siemens Healthineers). Scan range included whole body from the top of the head through the feet. The 18F-FDG-tracer dosing was adapted for body weight (3MBq 18F-FDG/kg). The visual evaluation of the PET was carried out using the Ultra-HD-3D mode reconstruction and the iterative 3D-mode reconstructions. The SUV measurements were conducted using the iterative 3D-mode reconstructions. The SUVmax was defined as maximal measured SUV value of the single most active intramedullary or extramedullary lesion on each scan using manually placed volumes of interest (VOI). Focal bone lesions were defined as FDG-avid focal areas within bone structure with tracer uptake intensity above median bone marrow activity (based on the prospective study by Bartel et al. [15]). The threshold of more than 3 focal lesions and the SUVmax cut-off value of 4 for survival analysis were based on prospective and larger retrospective studies of 18F-FDG-PET/CT prognostic role in first-line therapy (Table 1) and IMWG 2017 consensus statement [4]. The presence of PET-positive EMD was also evaluated. EMD was defined as FDG-avid lesion that was not contiguous to bone on CT-scan and arose in soft tissue.

Statistics

We used commercial software (SPSS ver. 23 by IBM Statistics) for statistical analysis. Kaplan–Meier survival analysis was performed to assess the time of OS and PFS with start-point at the moment of the PET/CT investigation. Survivors were censored at the time of the last clinical contact. In survival analysis, an event was defined as death of any reason for OS, and as next disease relapse or death for PFS. Log-rank tests were used to assess statistical significance. A significance level of p-value was set at 0.05. Besides PET parameters, clinical findings and laboratory results, as well as number of prior treatment lines, were evaluated as risk factors.

Results

Patient characteristics

Among 36 relapsed patients, 20 were male and 16 female (Table 2). Median age was 60 years (range 44–76). Seventeen patients previously underwent single autologous SCT, 12 patients received two autologous SCTs, and 1 patient had three autologous SCTs. Six patients underwent allogeneic SCT, five of them after previous autologous transplantation. Thirteen (36%) patients experienced first disease relapse, 9 (25%) had a second, 8 (22%) had a third, and 6 (17%) patients suffered a fourth or fifth relapse. With a median follow-up of 37.9 months, the median OS in the entire study population was 16.3 (SE 17.7) months. Median time of PFS of was 6.0 (SE 3.6) months. Five of 36 patients (14%) were lost to follow-up before reaching next progress or death.

PET parameters

At the time of relapse, 26/36 (72.2%) of our patients had a positive PET scan. Among those, 14/26 patients had only intramedullary disease, while 10/26 had both intra- and extramedullary FDG-avid lesions. Two patients had EMD only. Ten patients had a negative PET scan with no FDG-avid lesions detectable.

Maximal SUV

The median OS of the 16 patients with an SUVmax < 4 was not reached, while in the 20 patients with an SUVmax > 4 the estimated median survival was 3.9 (SE 8.01) months (p = 0.014, Fig. 1a). Using established first-line parameters, a significantly shorter time of PFS was observed in 20 patients with SUVmax > 4 (median of 4 (SE 0.17) vs. 15 (SE 4) months, p = 0.002) as well (Fig. 2a).

Fig. 1
figure 1

Kaplan–Meier analysis of prognostic impact of PET-CT parameters for overall survival. a According to maximum SUV over 4. b According to number of focal lesions, threshold over 3. c According to presence of EMD

Full size image
Fig. 2
figure 2

Kaplan–Meier analysis of prognostic impact of PET-CT parameters for progression free survival. a According to maximum SUV over 4. b According to number of focal lesions, threshold over 3. c According to presence of EMD

Full size image

Focal bone lesions

The presence of a single PET-positive focal lesion failed to achieve significance for OS in our population (p = 0.1). Looking at the established number of FL in first-line patients, the median survival of the 15 patients with more than 3 FLs was significantly shorter (3.8 months, SE 4.7) than in 21 patients with 3 or less FLs (median not reached, p = 0.003, Fig. 1b).

Similarly, the presence of a single PET-positive focal lesion was not a significant predictor for shorter PFS (p = 0.08), while the threshold of > 3 focal lesions achieved significance as negative predictor for PFS (3.5 (SE 0.14) vs. 15 (SE 3.8) months, p < 0.001) (Fig. 2b). Interestingly, the presence of > 3 focal bone lesions on CT alone, without PET assessment of lesion activity, had no statistical significance as a negative predictor of OS or PFS (Table 3).

Table 3 The results of statistical analysis
Full size table

Extramedullary disease

FDG-avid EMD was present in 12/36 (33.3%) of our patients. In patients with EMD the median OS was 3.63 (SE 2.2) months, compared to not reached in patients with no EMD on PET/CT (p = 0.004, Fig. 1c). In patients with EMD the median PFS was 3.5 (SE 0.4) months, compared to 11 (SE 1.8) months in patients with no EMD on PET-CT (p = 0.006, Fig. 2c). In 9/12 cases the SUVmax was located in the extramedullary lesion and in 8 of those 9 cases it exceeded the threshold of 4. In 8 of 12 patients radiation therapy of PET-positive extramedullary lesions was initiated, demonstrating relevance for therapy planning.

Laboratory and clinical parameters

Laboratory parameters were assessed at the time of PET/CT investigation (Table 3). Thirteen patients had ISS stage II and four had stage III. The OS was significantly shorter in patients with higher ISS stage (3.9 and 13.6 months vs. not reached, p = 0.016, Fig. 3a). The ISS stage had somewhat weaker, but still significant negative predictive value for PFS as well (p = 0.046, Fig. 3b). In 15/36 patients cytogenetic analysis was not available, precluding determination of disease stage according to the revised ISS (R-ISS). The role of elevated LDH (> 250 U/l), as an element of R-ISS, was assessed separately. LDH was elevated in 20/36 patients and was predictive of significantly shorter OS (6 months vs. not reached, p = 0.006, Fig. 3c) and PFS (4 vs. 15 months, p = 0.003, Fig. 3d). Patients relapsing after third-line or later line of therapy had significantly shorter PFS (p = 0.001) and showed a strong trend to lower OS (p = 0.069).

Fig. 3
figure 3

Kaplan–Meier analysis of prognostic impact of serum LDH levels and ISS-Stage for OS and PFS. a According to ISS—Stage (OS), b according to ISS—Stage (PFS), c according to LDH, threshold > 250 U/l (OS), d according to LDH, threshold > 250 U/l (PFS)

Full size image

Combining PET and ISS

To identify the patients with the highest risk, we evaluated the combined value of PET and ISS. We selected the cases in which both higher ISS stage (II or III) and at least one of the predictive PET parameters (EMD, nFL > 3, or SUVmax > 4) was present. These “high-risk” patients (n = 11) had significantly shorter OS than the rest of the population (3.7 (SE 0.2) months vs. not reached, p < 0.001), with minimal variability. The PFS was significantly shorter in these patients, too (3.63 (SE 0.14) vs. 11.7 (SE 2.2), p < 0.001). In contrast, the “low-risk” patients with ISS stage I and none of the predictive PET parameters survived significantly longer (OS: not reached vs. 13.6 months, p = 0.028; PFS: 20.7 vs. 5.9 months, p = 0.007). The comparison of “high-risk” versus “low-risk” patients versus others in a single model is demonstrated in Fig. 4.

Fig. 4
figure 4

Kaplan–Meier analysis of combined prognostic impact of ISS and PET-Parameters. a according to OS, b according to PFS. *negative PET—none of established predictive PET-parameters (SUVmax > 4, nFLs > 3, EMD) present. **positive PET—at least one of predictive PET-parameters (SUVmax > 4, nFLs > 3, EMD) present

Full size image

Discussion

Evaluation of the prognostic role of PET/CT in the first line setting of myeloma treatment identified useful thresholds for certain parameters. For instance, Bartel et al. [15] showed in a large prospective study of 239 patients that the presence of more than 3 FDG-avid focal bone lesions at baseline was an independent prognostic factor for OS and event-free survival (EFS), while the presence of FDG-avid EMD and a SUVmax > 3.9 were significant on monovariate analysis. These findings were confirmed in a prospective study of 192 patients by Zamagni et al. in 2011. Presence of FDG-avid EMD and SUVmax > 4.2 at baseline were identified as independent negative prognostic factors for OS and PFS [8]. The presence of more than 3 PET-positive FLs at day 7 of induction therapy was an independent negative prognostic parameter for OS and PFS in the large prospective study of 302 patients by Usmani et al. [11]. In 2015, Zamagni et al. demonstrated the negative prognostic role of baseline SUVmax > 4.2 for PFS and OS in a retrospective evaluation of 282 patients [9]. Patriarca et al. [10] demonstrated the negative prognostic role of SUVmax > 4.2 in 67 patients prior to autologous SCT. Davies et al. [14] corroborated the presence of more than 3 PET-positive FLs at baseline as a negative prognostic parameter for OS and PFS in a retrospective evaluation of 596 cases. Wang et al. [16] demonstrated the negative prognostic role of EMD present at baseline for both OS and PFS. The IMWG 2017 consensus statement proposed a threshold of SUV > 4 for defining significant PET-positivity [4].

In comparison, retrospective analyses in patients with relapse after SCT identified significantly higher prognostic thresholds. A study of 40 patients by Jamet et al. selected relapsed patients after high-dose chemotherapy and SCT. A SUVmax > 15.9 and more than 13 FLs (or > 6 FLs in appendicular skeleton) were identified as negative prognostic predictors for PFS [19]. Earlier, Lapa et al. [18] had performed a retrospective analysis of 37 patients suffering relapse after at least one autologous and/or allogeneic SCT. A SUVmax > 18.57 was predictive of shorter time to progression (TTP), while more than 10 FLs and the presence of EMD predicted both shorter TTP and OS. In both studies, patients were heavily pre-treated prior to relapse, which is true for our study as well.

We used a lower threshold of SUVmax > 4, based on the results of larger first line studies. In our relapse population, the SUVmax > 4 had a negative prognostic impact on both OS and PFS. A similar trend was seen regarding cut-offs for active FLs. Multiple pre-SCT studies of patients in first-line therapy identified a threshold of more than 3 FLs as a negative predictor for OS and/or PFS [8, 9, 14, 15, 17]. Similar to SUVmax, the threshold number of FLs with negative prognostic power was much higher (6 or 10 FLs) in retrospective studies of relapsed MM [18, 19]. However, the results of our analysis support the prognostic significance of the lower cut-off value of FL > 3 for OS and PFS in patients with relapsed MM. Furthermore, the presence of PET-positive EMD was a significant negative predictor of OS and PFS, in accordance with the results of multiple first line studies, and as demonstrated by Lapa et al. [18] for relapsed myeloma.

While the comparison with previous relapse studies is hampered by small study populations and variability of pretreatment intensity, our findings still provide a basis for comparing prognostic thresholds of PET/CT parameters in the first-line and relapse settings.

ISS stage is the most widely used tool for prognostic assessment and demonstrated prognostic significance for OS and PFS in our patient population. Combining ISS and PET-parameters allowed us to reliably identify patients with particularly short survival. We demonstrated that PET is an independent prognostic parameter, adding prognostic information to ISS stage. The combined use of PET thresholds and clinical parameters may enhance PET-based prognostic evaluation in the first-line and relapse setting.

Our analysis has certain limitations. Similar to other studies of PET/CT in MM relapse after SCT, patients in our study were heterogeneous regarding their pretreatment history. 23 of 36 patients (63.8%) had received at least 2 lines of therapy before current relapse, and 18 of 36 (50%) had received more than one SCT. Since patient selection for our retrospective analysis covered a 7-year period (2012–2019), treatment protocols varied substantially. In addition, small patient numbers render multivariate models less informative and put more emphasis on univariate analysis. Still, the established clinical prognostic parameters, including elevated LDH and higher ISS proved themselves as significant negative predictors for both OS and PFS in our relatively small study population. The applicability of established clinical prognostic parameters suggests that our patient cohort is representative of a wider population of patients with MM and supports the conclusions drawn from this study population.

Our results could contribute to the management of MM at relapse after SCT. Data up to now only showed that very high thresholds for PET-parameters, such as SUVmax of over 15 or 18 and nFLs of over 6 or 10 are prognostically relevant in relapse setting [18, 19]. We show that lower thresholds of FLs > 3 or SUVmax > 4 already identify patients with high risk and therefore should lead physicists to treat these patients more aggressively. The presence of extramedullary disease was a significant negative prognostic parameter both in our population and in previous studies.

Conclusions

We show that the same PET/CT parameter thresholds that are used in the first-line setting can also stratify patients with relapsed MM and identify those with particularly poor prognosis. Our findings demonstrate the prognostic yield of PET/CT in MM relapse independent of ISS stage and suggest that the same thresholds for PET prognostic parameters can be used at baseline and at later stages in the course of disease.Patients with relapsed MM who show unfavourable prognostic PET/CT parameters may survive only few months. The use of 18F-FDG-PET/CT, especially in combination with established clinical parameters, may thus help to select high-risk patients with MM relapse who may benefit from aggressive treatment approaches.