High-SUVmax on FDG-PET indicates pleomorphic subtype in epithelioid malignant pleural mesothelioma: Supportive evidence to reclassify pleomorphic as non-epithelioid histology

Abstract

Background

We have recently proposed to reclassify the pleomorphic subtype of epithelioid malignant pleural mesothelioma (MPM) as non-epithelioid (biphasic/sarcomatoid) histology due to its similarly poor prognosis. We sought to investigate whether preoperative maximum standardized uptake value (SUVmax) on ¹⁸F-fluorodeoxyglucose (FDG) positron emission tomography (PET) correlates with histologic subtype in MPM.

Methods

Clinical data was collected for 78 patients with MPM who underwent preoperative FDG-PET. We retrospectively classified the epithelioid tumors into five subtypes: trabecular, tubulopapillary, micropapillary, solid and pleomorphic. Tumors were categorized by SUVmax into two groups: low (<10.0) and high (≥10.0).

Results

The median overall survival of epithelioid tumors with high-SUVmax (n=12) was significantly shorter (7.1 months) than that of epithelioid tumors with low-SUVmax (n=54, 18.9 months, p<0.001) and comparable to non-epithelioid tumors (n=12, 7.2 months). Epithelioid tumors with pleomorphic subtype (n=9) had marginally higher SUVmax (mean±SD: 10.6±5.9) than epithelioid non-pleomorphic subtype (n=57, 6.5±3.2, p=0.050), and were comparable to that of non-epithelioid tumors (n=12, 9.1±4.8). Among the epithelioid tumors with high-SUVmax (n=12), 50% (n=6) showed pleomorphic subtype. In contrast, among epithelioid tumors with low-SUVmax (n=54), 6% (n=3) showed epithelioid pleomorphic subtypes (p=0.001). A positive correlation between mitotic count and SUVmax was observed (r=0.30, p=0.010).

Conclusions

Pleomorphic subtype of epithelioid MPM showed higher SUVmax than epithelioid non-pleomorphic subtype and was similar to non-epithelioid histology. Preoperative SUVmax on FDG-PET in epithelioid MPM can indicate patients with pleomorphic subtype with poor prognosis, supporting their reclassification as non-epithelioid.

INTRODUCTION

Diffuse malignant pleural mesothelioma (MPM) is an uncommon but aggressive tumor with median survival of 9 to 12 months despite multimodal therapy (surgery, chemotherapy and radiation therapy).¹ Histology and TNM (tumor, node, metastases) stage are the only standard predictors of survival.^2-4 While epithelioid MPM has a better prognosis than non-epithelioid (biphasic and sarcomatoid) tumors, prognosis within epithelioid histology is variable. We have recently reported the prognostic utility of histologic subtyping in epithelioid MPM and proposed that the pleomorphic subtype should be reclassified as non-epithelioid histology due to similar clinical outcomes.⁵ However, to the best of our knowledge, the biological reasons for this similarity in prognosis remain unexplored.

¹⁸F-fluorodeoxyglucose (FDG) positron emission tomography (PET) is a standard radiographic tool in clinical practice that assesses the metabolic activity of tumor cells.^6-8 In addition to facilitating prognosis, maximum standardized uptake value (SUVmax) on FDG-PET reflects histology in lung cancer. SUVmax is significantly lower in lung adenocarcinoma than squamous cell carcinoma.^9-11 To investigate the biology of the pleomorphic subtype in epithelioid MPM, our aim in this study was to determine the correlation between preoperative SUVmax and histologic subtypes of epithelioid MPM. In addition, we investigated the correlation between SUVmax and tumour proliferation based on mitotic count.

MATERIALS AND METHODS

Patients

Tumor slides were available for 148 patients who received a diagnosis of MPM between 1998 and 2009 at Memorial Sloan-Kettering Cancer Center (MSKCC). Of these, 78 patients underwent FDG-PET before surgical resection. Fifty of 78 patients (64%) underwent PET scans at MSKCC. Seventy-one patients (91%) were not treated with any chemotherapy prior to PET scans. Clinical information was collected through a database maintained by the Thoracic Service, Department of Surgery at MSKCC. Institutional review board approval was obtained at MSKCC before the study began. There were 66 patients with epithelioid tumors and 12 with non-epithelioid tumors (6 biphasic and 6 sarcomatoid). Clinical variables recorded in the prospectively maintained database included age, gender, laterality, TNM stage and surgical procedure. TNM staging was based on the reported imaging findings, the surgeon’s intraoperative findings and the pathologic evaluation of the resected specimens using the sixth edition of the American Joint Commission on Cancer Staging Manual.¹² All patients were followed until date of death or last follow-up.

Pathologic diagnosis was based on standard histologic, histochemical and immunohistochemical criteria.^13-15 As a positive marker of immunohistochemistry for MPM, we used standard immunohistochemical markers including calretinin, WT-1, cytokeratin 5/6 and D2-40. As negative markers for MPM, we used carcinoembryonic antigen, CD15, B72.3, BerEP4 and thyroid transcription factor-1. In addition, pathologic diagnosis was correlated with gross distribution of the tumor and absence of an intrapulmonary lesion on radiologic imaging.

Technique of FDG-PET

The following technique was used for PET scans performed at MSKCC. Patients received 10–15 mCi (370–555 MBq) of FDG intravenously. Patients were instructed to fast ≥6 hours before injection; plasma glucose levels were measured prior to imaging. Approximately 60 minutes after injection, torso images were acquired with either GE Advance (GE Medical Systems, Waukesha, WI) or HR plus (Siemens/CTI, Knoxville, TN) PET scanners. Beginning in November 2001, studies were also acquired on hybrid PET/CT (computed tomography) imaging systems, including the Biograph (Siemens/CTI, Nashville, TN) and Discovery LS (GE Medical Systems, Waukesha, WI). The Biograph data was acquired in 3D mode. All the other scanners used 2D PET image acquisition. Discovery LS incorporates a PET Advance tomograph, and Biograph incorporates an HR plus PET tomograph. For PET/CT, a low-dose CT scan was acquired first, to allow for PET attenuation correction and anatomic localization of PET abnormalities. Each PET dataset was reconstructed for image display using iterative algorithms, with and without attenuation correction. Experienced radiologists with specific expertise in nuclear medicine interpreted PET imagery at the time of diagnosis. Uptake of FDG by tumor was quantified by PET region-of-interest analysis with the SUVmax. SUV was calculated as:

$$\mathrm{SUV}=\frac{\left(\text{decay-corrected activity}\right[\mathrm{kBq}]∕\text{tissue volume}[\mathrm{ml}\left]\right)}{\left(\text{injected-FDG activity}\right[\mathrm{kBq}]∕\text{body weight}[\mathrm{g}\left]\right)}$$

Histologic evaluation

All available hematoxylin and eosin–stained slides (median 7, range 1–43 slides/case) of epithelioid MPM lesions were reviewed by a single pathologist (KK) for the purpose of this study using an Olympus BX51 microscope (Olympus, Tokyo, Japan) with a standard eyepiece of 22 mm diameter; problem cases were reviewed by two pathologists (WDT and KK). Histologic classification for epithelioid MPM was done according to the 2004 World Health Organization criteria (<10% sarcomatoid component).¹⁵ Epithelioid MPM can be comprised of one or more of five histologic patterns,⁵ which were recorded in 5% increments: (1) trabecular, (2) tubulopapillary, (3) micropapillary, (4) solid and (5) pleomorphic. Tumors were classified as pleomorphic subtype when cytologic pleomorphism comprised at least 10% of the tumor.⁵ The remaining tumors were classified according to the predominant histologic patterns.

Mitoses were evaluated using high-power-field (HPF) at ×400 magnification (0.237 mm² field of view) in the 50 HPF areas with the highest mitotic activities^16-19 and counted as an average of mitotic figures per 10 HPF. In the cases in which only small areas of viable tumor were available for review, the best attempt was made to assess the equivalent of 10 full HPF of viable tumor for mitosis counting.¹⁷

We also recorded the following histological factors: presence of lymphatic or vascular invasion, necrosis (%), fibrosis (%) and myxoid change (%).

Statistical analysis

Associations between clinicopathologic variables and histologic findings were analyzed using a Fisher exact test for categorical variables and Wilcoxon test for continuous variables. Overall survival (OS) following surgery was estimated using the Kaplan-Meier method, with patients censored if they were alive at the time of last follow-up. An analysis of time to recurrence (TTR) was restricted to patients who underwent surgery that was deemed to be a complete resection. Non-parametric group comparisons were performed using log-rank test. All p values were based on two-tailed statistical analysis and a p value of <0.05 was considered to indicate statistical significance. All analyses were performed using SAS statistical software (version 9.2; SAS Institute Inc, Cary, NC).

RESULTS

Clinicopathologic demographics and their associations with OS

The clinicopathologic profile of 66 patients with epithelioid MPM is outlined in table 1. Median age was 63 (range 29–81); and 65% (n=43) were males. The tumor involved the left pleura in 50% (n=33) of the cases. Three patients (5%) were stage I, 16 (24%) were stage II, 33 (50%) were stage III and 14 (21%) were stage IV. Six patients (9%) were treated with chemotherapy prior to PET scan. Sixteen patients (24%) underwent pleurodesis prior to PET scans. Eleven of them (67%) underwent pleurodesis more than one month prior to PET. By surgical procedure, 37 (56%) underwent extrapleural pneumonectomy, 21 (32%) underwent pleurectomy-decortication and the remaining 8 (12%) had other procedures (3 biopsies, 4 exploratory thoracotomies and 1 palliative pleurectomy). Lymphatic invasion was detected in 42% (n=28) and vascular invasion in 20% (n=13). By histologic subtype, 9 tumors (14%) were pleomorphic and 57 (86%) were non-pleomorphic. Within the 6 patients with epithelioid MPM who underwent chemotherapy prior to PET scan, one tumor was pleomorphic and 5 were non-pleomorphic.

Table 1.

SUVmax associations with clinicopathologic factors in patients with epithelioid malignant pleural mesothelioma

Variables	n	SUVmax (mean±SD)	p value
All patients	66	7.1±3.9
Age			0.454
≤65	26	7.5±4.3
>65	40	6.3±3.1
Gender			0.086
Female	23	6.6±4.8
Male	43	7.3±3.3
Laterality			0.640
Left	33	7.5±4.5
Right	33	6.6±3.1
T stage			0.530
T1	6	6.1±3.8
T2	24	6.6±3.7
T3	25	8.1±4.5
T4	11	6.1±2.6
N stage			0.240
N0	42	6.8±4.1
N1	4	7.6±2.5
N2	20	7.5±3.7
Stage			0.490
I	3	6.8±5.4
II	16	6.5±4.1
III	33	7.6±4.2
IV	14	6.5±2.4
Chemotherapy prior to PET			0.867
Yes	6	7.1±3.5
No	60	7.1±4.0
Pleurodesis prior to PET			0.805
Yes	16	6.8±3.7
No	50	7.1±4.0
Lymphatic invasion			0.750
Absence	38	6.9±3.9
Presence	28	7.2±3.9
Vascular invasion			0.054
Absence	53	6.6±3.6
Presence	13	9.1±4.5
Fibrosis			0.330
<50%	43	7.6±4.4
≥50%	23	6.1±2.5
Necrosis			0.042
<10%	59	6.6±3.4
≥10%	7	10.8±5.8
Myxoid			0.820
<50%	61	7.0±3.9
≥50%	5	7.2±3.7
Histologic subtype			0.050
Non-pleomorphic	57	6.5±3.2
Pleomorphic	9	10.6±5.9

SUVmax, maximum standard uptake value, SD, standard deviation

Median OS was 16.3 months. On univariate analyses, necrosis ≥10% was associated with shorter OS (p=0.002). No other clinicopathologic factors were significantly associated with OS.

SUVmax and its association with OS and TTR

Among the epithelioid MPM lesions, SUVmax ranged from 1.7 to 21.0 (median 6.3, mean±SD 7.1±3.9). Tumors were classified into 2 groups by SUVmax as previously reported: low-SUVmax <10 and high-SUVmax ≥10.^20,21 The median OS of patients with epithelioid tumors with high-SUVmax (n=12) was significantly shorter (7.1 months) than that of patients with epithelioid tumors with low-SUVmax (n=54, 18.9 months, p<0.001), and comparable to patients with non-epithelioid tumors (n=12, 7.2 months), as shown in figure 1A.

Figure 1.

Overall survival (OS) and time to recurrence (TTR) by maximum standardized uptake value (SUVmax).

(A) The median OS of patients with epithelioid tumors with high-SUVmax (n=12) was significantly shorter (7.1 months) than those with epithelioid tumors with low-SUVmax (n=54, 18.9 months) and comparable to patients with non-epithelioid tumors (n=12, 7.2 months). (B) In 44 patients with completely resected epithelioid tumors, the median TTR for those with high-SUVmax (n=7) was significantly shorter (14.4 months) than those with low-SUVmax (n=37, 21.9 months).

We have recently proposed to reclassify the pleomorphic subtype of epithelioid MPM as non-epithelioid histology,⁵ so we repeated our analysis of SUVmax in a cohort of patients with epithelioid MPM, excluding the 9 pleomorphic cases. Among these 57 epithelioid cases, those with epithelioid tumors with high-SUVmax (n=6) had a shorter median OS (5.6 months) than those with low-SUVmax (n=51, 19.4 months, p=0.005).

Recurrence was observed in 19 of the 44 patients with epithelioid tumors that underwent complete resection; the median TTR for patients with epithelioid tumors with high-SUVmax (n=7) was significantly shorter (14.4 months) than for low-SUVmax (n=37, 21.9 months, p=0.047), as shown in figure 1B.

Association between SUVmax and histology

Within the epithelioid MPM, tumors with pleomorphic subtype (n=9) had the highest SUVmax (mean±SD: 10.6±5.9), followed by solid (n=20, 6.5±2.9), micropapillary (n=10, 6.1±2.8), tubulopapillary (n=18, 6.7±4.0) and trabecular (n=9, 6.7±3.0) (figure 2A). Epithelioid tumor with pleomorphic subtype had marginally higher SUVmax than epithelioid non-pleomorphic subtype (n=57, 6.5±3.2, p=0.050). Non-epithelioid MPM (n=12) appeared to have higher SUVmax (9.1±4.8) compared to epithelioid tumors (n=66, 7.1±3.9), although the difference was not significant (p=0.160). However, SUVmax of non-epithelioid MPM was similar to epithelioid tumor with pleomorphic subtype and significantly higher than epithelioid non-pleomorphic subtypes (p=0.046) (figure 2B). Among the epithelioid tumors with high-SUVmax (n=12), 50% (n=6) showed pleomorphic subtype. In contrast, among epithelioid tumors with low-SUVmax (n=54), 6% (n=3) showed pleomorphic subtypes (p=0.001).

Figure 2.

Association between maximum standardized uptake value (SUVmax) and histology.

(A) Within the epithelioid tumors, pleomorphic subtype (n=9) had the highest SUVmax (mean±SD: 10.6±5.9), followed by solid (n=20, 6.5±2.9), micropapillary (n=10, 6.1±2.8), tubulopapillary (n=18, 6.7±4.0), and trabecular (n=9, 6.7±3.0). (B) SUVmax of non-epithelioid tumors (n=12, 9.1±4.8) was similar to pleomorphic subtype and significantly higher than epithelioid non-pleomorphic subtype (n=57, 6.5±3.2).

Association between SUVmax and clinicopathologic factors

Associations between SUVmax and clinicopathologic factors in patients with epithelioid MPM are outlined in table 1. Among these factors, necrosis ≥10% was significantly associated with higher SUVmax (p=0.042). When examining mitotic count as a continuous variable, a moderate correlation between mitotic count and SUVmax was observed (r=0.30, p=0.010). Male gender (p=0.086) and vascular invasion (p=0.054) showed a tendency to have higher SUVmax. However, there were no associations between SUVmax and age, disease laterality, TNM stage, chemotherapy prior to PET scan, pleurodesis prior to PET scan, lymphatic invasion, fibrosis or myxoid change (greater than or equal to 50%).

SUVmax of the patients who underwent PET scan at MSKCC and its association with OS and histology

Among the epithelioid MPM patients who underwent PET scan at MSKCC (n=43), SUVmax ranged from 2.3 to 16.5 (median 6.5, mean±SD: 7.2±3.4). Among the patients who underwent PET scan using Ge-68 transmission rods (GE Medical Systems) without CT (n=14), SUVmax ranged 3.3 to 15.0 (median 6.6, mean±SD: 7.2±2.9). Among the patients who underwent PET/CT scan (n=29), SUVmax ranged 2.3 to 13.4 (n=14, median 4.7, mean±SD: 5.6±3.0) on Biograph and 3.6 to 16.5 (n=15, median 8.3, mean±SD: 8.6±3.6) on Discovery LS.

Within the patients who underwent PET scan at MSKCC (n=50), which comprised two-third of total cohort, the median OS of patients with epithelioid tumors with high-SUVmax (n=7) was significantly shorter (8.9 months) than that of patients with epithelioid tumors with low-SUVmax (n=36, 19.4 months, p<0.001), and comparable to patients with non-epithelioid tumors (n=7, 6.9 months). SUVmax of tumors with epithelioid non-pleomorphic subtypes (n=37, mean±SD: 6.9±2.9) was lower than epithelioid tumors with pleomorphic subtype (n=6, 8.8±5.5) and non-epithelioid tumors (n=7, 10.1±5.4) although the small sample size did not allow to perform statistical analysis. Among the epithelioid tumors with high-SUVmax (n=7), 43% (n=3) showed pleomorphic subtype. In contrast, among epithelioid tumors with low-SUVmax (n=36), 8% (n=3) showed pleomorphic subtypes.

DISCUSSION

Current preoperative therapeutic decisions of MPM are based on histologic type and the TNM stage. To account for the histological heterogeneity among epithelioid MPM, we have recently assessed the prognostic significance of five histologic subtypes and observed that the pleomorphic subtype resembles more closely the clinical outcome of non-epithelioid (biphasic and sarcomatoid) histology.⁵ To gain further insight into the biology of the pleomorphic subtype, we investigated whether the SUVmax reflects the histologic subtypes in epithelioid MPM. We demonstrated that high-SUVmax was not only associated with disease recurrence and OS, but also with increased proportion of pleomorphic subtype and proliferative activity based on mitotic count in patients with epithelioid MPM.

Pleomorphic subtype is defined as having >10% of the tumor demonstrating pleomorphism.⁵ Accurately identifying these patients prior to deciding on appropriate therapeutic management is difficult because the small proportion of pleomorphism may not be obvious on small diagnostic biopsy samples. Yet, given their poor prognosis (median OS=8.1 months),⁵ the ability to recognize this population is critical to clinical decision-making. In the current study, we observed that the 9 epithelioid tumors with pleomorphic subtype were characterized by having higher SUVmax (mean SUVmax=10.6) compared to epithelioid non-pleomorphic subtypes (mean SUVmax=6.1-6.7) (figure 2A). Among the patients with epithelioid MPM with SUVmax >10, half had the pleomorphic subtype, while for those with SUVmax <10, the majority (94%) had the non-pleomorphic subtype. Though it is difficult to precisely correlate the pleomorphic area on microscopic examination to the site of SUVmax on FDG-PET, the association we observed may help in preoperative identification of the pleomorphic subtype.

Furthermore, we observed a similarity in the SUVmax of pleomorphic subtype (mean SUVmax=10.6) and the non-epithelioid histology (mean SUVmax=9.1) (figure 2B). In our previous report we demonstrated the resemblance in outcomes between the pleomorphic subtype and the non-epithelioid histology (the pleomorphic subtype experienced median OS of 8.1 months compared to 7.0 months and 3.0 months for biphasic and sarcomatoid, respectively).⁵ Our current observation provides radiographic evidence that the pleomorphic subtype more closely resembles the non-epithelioid MPM than the epithelioid histology and further strengthens our proposal to reclassify the pleomorphic subtype under the non-epithelioid histology.

High FDG uptake has been shown to correlate with decreased survival in patients with MPM.^20-23 However, these studies were heterogeneous in morphology because they included non-epithelioid tumors. Therefore, the prognostic utility of FDG uptake was unclear in a uniform cohort comprised of epithelioid histology. Within our cohort of 66 patients with epithelioid MPM, a cutoff of SUVmax 10 significantly stratified OS, and this finding remained significant even after excluding patients with pleomorphic subtype.

In our study, 24% of epithelioid MPM patients underwent pleurodesis prior to PET scans. While the prolonged, marked hypermetabolic pleural activity associated with pleurodesis potentially may limit PET scan evaluation, the pleural SUVmax in the epithelioid MPM patients who underwent pleurodesis prior to PET scan showed no significant difference from the patients with no prior pleurodesis.

One limitation of the current study is that the PET scans were performed at multiple locations. Due to the rarity of MPM, it is difficult to obtain a large number of scans at a single location. Nevertheless, two-thirds of the scans in this study were performed at the same location. In order to validate our finding with a more uniform group of patients, we performed the subsequent analysis on the patients who underwent PET scan at MSKCC. In this group, SUVmax remained significant in stratifying OS, and the epithelioid tumor with pleomorphic subtype demonstrated higher SUVmax than epithelioid non-pleomorphic subtypes. However, limitation of our study is lack of standardization. But as MPM is a rare disease, even at a tertiary referral center like MSKCC, many of the patients are often referred following an initial surgical procedure or pleurodesis.

Another limitation of our study is that the PET scans were performed by 4 different scanners at MSKCC, and the SUV differences were observed by the different scanners, attenuation corrections, and acquisition models. Although the use of 4 different scanners is a potential limitation of this study, our correlations are significant despite the limitations of our study – differences in CT versus rod-source attenuation correction²⁴ and differences in acquisition of data in 2D or 3D models.²⁵ In a more positive perspective, however, the fact that we were able to detect statistically significant associations between SUVmax versus OS and histology, despite using different scanners, may suggest that (1) our results are generalizable to other medical centers using different scanners than ours and (2) the true correlation between SUVmax versus OS and histology may be under-estimated by our results, presuming that using different scanners will weaken such correlation rather than strengthen it. The previous report suggested the SUV differences by the different acquisition models (2D or 3D).²⁵ In our study, the Biograph data was acquired in 3D mode, and all the other scanners used 2D PET image acquisition. Again, although these technical differences may weaken the ability to detect correlations that we reported in this study, we detected the correlations despite these technical limitations, not because of them.

While future studies should normalize SUVmax by different scanners, our findings confirm its prognostic value. In a recent report, Nowak et al proposed a prognostic model that incorporates total glycolytic volume (TGV) on FDG-PET to account for the tumor volume in addition to the metabolic activity. Interestingly, in their cohort of 89 patients, TGV was prognostic while SUVmax was not.²⁶ We did not record TGV in the current study.

In addition to prognosis, SUVmax correlates with a higher proliferation index in lung cancer.^7,27 However, the correlation between SUVmax and proliferative factor, such as Ki-67 labeling index and mitotic count, has not been investigated in MPM. In our study, higher SUVmax showed a correlation with higher mitotic count, demonstrating that FDG uptake by MPM tumor cells reflects proliferative activity. We also demonstrated that necrosis correlates with higher SUVmax. These findings confirm that SUVmax reflects tumor biology in addition to having prognostic value.

In conclusion, we have demonstrated in this study a radiologic-pathologic correlation in epithelioid MPM through comparison using a noninvasive imaging modality that measures tumor metabolism (SUVmax on FDG-PET) and histologic characterization on microscopic examination (histologic subtype). We observed that (1) high-SUVmax correlates with pleomorphic subtype in epithelioid MPM, thus providing radiologic resemblance of the pleomorphic subtype with the non-epithelioid tumors, and (2) epithelioid MPM with high-SUVmax should be considered as having a poorer prognosis, similar to non-epithelioid tumors. As this is a retrospective study, these findings should be validated prospectively. Nonetheless, our findings emphasize the importance of a multidisciplinary approach in both predicting patient outcomes and better understanding a heterogeneous disease such as MPM.