Incidence and characteristics of pseudoprogression in IDH-mutant high-grade gliomas: A POLA network study

Abstract

Background

Incidence and characteristics of pseudoprogression in isocitrate dehydrogenase-mutant high-grade gliomas (IDHmt HGG) remain to be specifically described.

Methods

We analyzed pseudoprogression characteristics and explored the possibility of pseudoprogression misdiagnosis in IDHmt HGG patients, treated with radiotherapy (RT) (with or without chemotherapy [CT]), included in the French POLA network. Pseudoprogression was analyzed in patients with MRI available for review (reference cohort, n = 200). Pseudoprogression misdiagnosis was estimated in this cohort and in an independent cohort (control cohort, n = 543) based on progression-free survival before and after first progression.

Results

In the reference cohort, 38 patients (19%) presented a pseudoprogression after a median time of 10.5 months after RT. Pseudoprogression characteristics were similar across IDHmt HGG subtypes. In most patients, it consisted of the appearance of one or several infracentimetric, asymptomatic, contrast-enhanced lesions occurring within 2 years after RT. The only factor associated with pseudoprogression occurrence was adjuvant PCV CT. Among patients considered as having a first true progression, 7 out of 41 (17%) in the reference cohort and 35 out of 203 (17%) in the control cohort were retrospectively suspected to have a misdiagnosed pseudoprogression. Patients with a misdiagnosed pseudoprogression were characterized by a time to event and an outcome similar to that of patients with a pseudoprogression but presented with larger and more symptomatic lesions.

Conclusion

In patients with an IDHmt HGG, pseudoprogression occurs later than in IDH-wildtype glioblastomas and seems not only frequent but also frequently misdiagnosed. Within the first 2 years after RT, the possibility of a pseudoprogression should be carefully considered.

Key Points.

• Pseudoprogression is frequent in isocitrate dehydrogenase (IDH)-mutant high-grade gliomas.

• The possibility of a pseudoprogression should be carefully considered within the first 2 years after the end of radiotherapy.

• Differentiating tumor progression from pseudoprogression can be difficult in patients presenting with large and symptomatic lesions.

Importance of the Study.

Characterizing pseudoprogression is important for patient management and the design of clinical trials. In the present study, based on a large series of patients prospectively included in the French POLA network, we show that pseudoprogression is frequent in isocitrate dehydrogenase-mutant high-grade gliomas (IDHmt HGG) and typically occurs within the first 2 years after radiotherapy (RT) completion. We also provide evidence that in addition to being frequent, pseudoprogression may be frequently misdiagnosed as true tumor progression when patients present with large and symptomatic lesions. Our results demonstrate the need of careful clinical and radiological evaluation of IDHmt HGG patients who present increased or new contrast-enhanced lesions within the first 2 years after RT with or without chemotherapy.

In neuro-oncology, pseudoprogression can be defined as radiographic changes mimicking tumor progression resolving or stabilizing without modifying therapy. It is an important issue in glioma patients treated with radiotherapy (RT) (with or without chemotherapy [CT]) since there is currently no accurate diagnostic method to distinguish pseudoprogression from true progression.^1–5 Although its pathophysiology is not fully understood, pseudoprogression likely corresponds to RT-induced inflammation and blood-brain barrier disruption.^3,6 Pseudoprogression was initially described in glioblastomas and considered to occur within the first 3 months after RT.^1,7–9 Subsequently, it was recognized that it could occur beyond 6-12 months after RT, and it was reported in most glioma subtypes, including lower-grade diffuse gliomas, circumscribed gliomas, and brainstem gliomas.^10–15 However, the frequency of occurrence and characteristics of pseudoprogression have not been specifically described in isocitrate dehydrogenase-mutant (IDHmt) high-grade gliomas (HGG). These gliomas consist of grade 3 IDHmt and 1p/19q codeleted oligodendrogliomas (O3), as well as grade 3 and grade 4 IDHmt astrocytomas (A3 and A4).¹⁶ IDHmt HGG have a much better prognosis than IDH-wildtype glioblastomas^17–19 and are characterized by a distinct gliomagenesis.^20–23 The identification and characterization of pseudoprogression in IDHmt HGG are important for patient care, but also because clinical trials are being conducted in patients with such tumors.²⁴ Herein, we analyzed the characteristics and frequency of occurrence of pseudoprogression in IDHmt HGG patients included in the French POLA network. Since the diagnosis of pseudoprogression can be difficult, we also aimed to explore the frequency of pseudoprogression misdiagnosis in these patients.

Methods

Patients

In France since 2008, there is a dedicated program to harmonize the management of de novo adult HGG with an oligodendroglial component (the Prise en charge des Oligodendrogliomes Anaplasiques [POLA; management of anaplastic oligodendrogliomas] network). This program has enabled the collection of a large series of IDHmt HGG for which clinical, radiological, histo-molecular, treatment characteristics, and outcome have been prospectively collected using a dedicated electronic case report form (e-CRF). The occurrence of a pseudoprogression, however, was not prospectively collected. For the purpose of the present study, we analyzed pseudoprogression characteristics and estimated pseudoprogression misdiagnosis in IDHmt HGG patients, treated with RT (with or without CT), included in the POLA network. Pseudoprogression was analyzed in patients with MRI available for review (reference cohort). Pseudoprogression misdiagnosis was estimated in this reference cohort and in the rest of the cohort (control cohort). Patients provided written informed consent for clinical data collection and genetic analysis according to national and POLA network policies. The design of the study was approved by the institutional review board and conducted according to the European ethical guidelines (MR004 n°22 5774).

Definition of Pseudoprogression and Estimation of the Rate of Pseudoprogression Misdiagnosis

Pseudoprogression was defined as the increase or the appearance of a contrast-enhanced lesion after RT (with or without CT) that disappeared or remained stable for at least 12 months without new oncological treatment; the use of corticosteroids was accepted. In order to estimate the frequency of patients with a pseudoprogression who could have been misdiagnosed as true progression, we analyzed overall survival (OS) and progression-free survival (PFS) before (PFS1) and after first progression (PFS2) in reference and control cohort, hypothesizing that patients with a misdiagnosed pseudoprogression would have a PFS2 longer than PFS1 and a prolonged survival. The PFS2/PFS1 ratio threshold suggestive of pseudoprogression was defined based on PFS before and after pseudoprogression in patients with a definite pseudoprogression in the reference cohort. The time periods used to define pseudoprogression and misdiagnosed pseudoprogression are summarized in Supplementary Figure S1.

Data Collection

Baseline clinical, histo-molecular, treatment characteristics, and outcome were extracted from the POLA network e-CRF. All medical and radiological files of patients included in the reference cohort were reviewed to determine whether patients had presented a pseudoprogression. The following characteristics were collected at the time of pseudoprogression and at the time of tumor progression: Karnofsky performance status, corticosteroid treatment initiation, the presence of new neurologic symptoms, time of onset after RT completion, time to resolution, pattern of contrast enhancement (nodular or ring-like), largest diameter of the contrast-enhanced lesion (<1 cm and ≥1 cm), and location of the contrast enhancement—classified as either lobar or deep-seated (corpus callosum, cingulate gyrus, basal ganglia, thalamus, brainstem, or periventricular location). When available, perfusion MRI and ¹⁸F-DOPA PET scan results were also reviewed. Relative cerebral blood volume and ¹⁸F-DOPA PET metabolism were classified as either increased or not according to the interpretation of the radiologist or the nuclear medicine specialist.

Statistical Analyses

Categorical variables were presented as frequencies and proportions, and continuous variables as medians and ranges. Comparison of quantitative variables was performed using the chi-squared test, Fischer’s exact test, or Fisher-Freeman-Halton test; for the quantitative variables, t test (two-tailed) or one-way ANOVA were used. OS was calculated from the date of surgery to that of last follow-up visit or death. PFS1 was calculated from the date of surgery to that of first progression or death; PFS2 was calculated from the date of first progression to that of second progression or death. The Kaplan-Meier method was used to estimate OS and PFS, and the log-rank test for comparisons of subgroups. When comparing the survival of patients with and without pseudoprogression, there may be a time bias since patients with pseudoprogression have to live long enough to meet the definition of pseudoprogression.¹⁰ To correct for this possible time bias, as done by West et al, patients without pseudoprogression were excluded from the analysis if they progressed (PFS analysis) or died (OS analysis) before the 95th percentile of the time to development of pseudoprogression.¹⁰ All statistical tests were two-sided with a significance level of .01 because of multiple comparisons. Statistical analyses were performed using SPSS Statistics for Windows, Version 23.0 (IBM Corp., Armonk, NY, USA).

Results

Between January 2008 and January 2020, a total of 2315 patients were included in the POLA network cohort; 743 of these had IDHmt HGG treated with RT (with or without CT). Among the latter, MRIs were available for review in 200 patients who were included in the reference cohort that was used to describe pseudoprogression characteristics and estimate the rate of pseudoprogression misdiagnosis. The 543 remaining patients constituted the control cohort. This cohort was used as an independent cohort to validate the estimation of pseudoprogression misdiagnosis in the reference cohort (Supplementary Figure S2).

Incidence and Characteristics of Pseudoprogression in the Reference Cohort

Among the 200 patients included in the reference cohort, 99 had an O3 (49%), 50 an A3 (25%), and 51 an A4 (26%). Their characteristics are summarized in Table 1.

Table 1.

Summary of Patient Characteristics at Diagnosis in the Reference Cohort

O3	A3	A4	Total	P value
Number (%)	99 (49)	50 (25)	51 (26)	200 (100)
Clinical characteristics at diagnosis
Sex ratio m/f, %	60/40	52/48	49/51	55/45	.41+
Median age, years (range)	48 (22-78)	37.5 (23-65)	38 (22-66)	42 (22-78)	<.001°
KPS, median	100	100	100	100	.12°
Seizures, n (%)	74 (75)	38 (76)	33 (65)	145 (72)	.35+
Radiological characteristics at diagnosis, n (%)
Frontal location	76 (77)	31 (62)	31 (61)	138 (69)	.06+
Contrast enhancement	71 (52)	20 (40)	38 (74)	129 (65)	<.001+
Mass effect	58 (59)	28 (56)	36 (71)	122 (61)	.25+
Calcifications	31/76 (41)	2/40 (4)	3/34 (10)	36/150 (24)	<.001#
Histo-molecular characteristics, n (%)
MVP	74 (74)	0 (0)	48 (94)	122 (61)	<.001+
Necrosis	22 (22)	0 (0)	13 (30)	35 (18)	<.001+
IDH mutation	99 (100)	50 (100)	51 (100)	200 (100)	1#
1p19q codeletion	99 (100)	0 (0)	0 (0)	99 (49)	<.001#
MGMT promoter methylation	33/42 (78)	17/24 (71)	9/13 (69)	59/79 (75)	.64+
CDKN2A homozygous deletion	5/98 (5)	0/55 (0)	11/41 (23)	16/194 (8)	<.001#
Treatment characteristics, n (%)
Type of surgery					.69+
Biopsy	22 (22)	8 (16)	7 (14)	37 (19)
GTR	21 (21)	10 (20)	13 (26)	44 (22)
STR	56 (57)	32 (64)	30 (60)	118 (59)
Postoperative treatment					<.001+
RT alone	22 (22)	2 (4)	3 (6)	27 (14)
RT + PCV	67 (68)	38 (76)	17 (33)	122 (61)
RT + Temozolomide	10 (10)	10 (20)	31 (61)	51 (25)
Outcome
Follow-up, median months (range)	75 (14-142)	56.5 (26-97)	64 (4-124)	65 (4-142)	<.001°
OS, median months (95% CI)	NRa	NRb	95 (78-112)	NRc	<.001^
With progression, n (%)	18 (18)	5 (10)	23 (45)	46 (23)	<.001+
5-year survival, %	94	96	68	87

Abbreviations: 95% CI, 95% confidence interval; A3, astrocytoma IDHmt grade 3; A4, astrocytoma IDHmt grade 4; m/f, male/female; CT, chemotherapy; GTR, gross total resection; KPS, Karnofsky performance status; MVP, microvascular proliferation; NR, not reached; O3, oligodendroglioma IDHmt and 1p/19q codeleted grade 3; OS, overall survival; PCV, Procarbazine, CCNU, Vincristine; RT, radiation therapy; STR, subtotal resection.

⁺Chi-squared test,

^#Fisher’s exact test or Fisher-Freeman-Halton exact test,

^{^}Log-rank test,

^°One-way ANOVA.

^aLast observation censored at 142 months (63% survival estimate).

^bLast observation censored at 97 months (55% survival estimate).

^cLast observation censored at 142 months (56% survival estimate).

After a median follow-up of 65 months, 38 patients (19%) presented a pseudoprogression. Pseudoprogression occurred after a median time of 10.5 months (range: 0-66 months). In 12 patients (32%) it occurred <6 months and in 26 (68%) patients ≥6 months after RT (Table 2); in most patients (n = 31, 82%) it occurred within 2 years after RT. In all but two patients (n = 36, 95%), pseudoprogression was not associated with new symptoms. It consisted of the appearance of one (n = 22, 58%) or several (n = 16, 42%) new contrast-enhanced lesions in 32 patients (84%), and in the increase of a preexisting contrast-enhanced lesion in 6 patients (16%; illustrated in Figure 1A). Most new contrast-enhanced lesion measured <1 cm (n = 24, 74%). Deep-seated location (including subependymal involvement) was found in 42% of cases. Twenty-six patients underwent a perfusion MRI, which demonstrated no significant relative cerebral blood volume (rCBV) increase in 24 patients (92%) and eight patients underwent a ¹⁸F-DOPA PET scan, which demonstrated no significant hypermetabolism in 7 patients (87%). New or increased contrast-enhanced lesion disappeared in 24 patients (63%) after a median time of 6 months (range: 2-30 months); it persisted in 14 patients (37%; range: 26-124 months). There was no significant difference between O3, A3, and A4 in terms of frequency, timing, or characteristics of pseudoprogression. Compared to patients who presented an early pseudoprogression (<6 months), there was a trend toward smaller contrast-enhanced lesion (67% vs 92%, respectively, P = .06) in those who presented a late pseudoprogression (≥6 months), and a trend toward deep location (8% vs 31%, respectively, P = .22).

Table 2.

Summary of the Characteristics of Patients Who Presented a Pseudoprogression in the Reference Cohort

O3	A3	A4	Total	P value
PsP, n/N (%)	18/99(18)	11/50 (22)	9/51 (18)	38/200 (19)	.82+
Timing of PsP
Median time after RT, months (range)	13.5 (1-66)	6 (0-31)	10 (1-33)	10.5 (0-66)	.34°
<6 months after RT, n (%)	5 (27)	4 (36)	3 (33)	12 (32)	.90#
6 months after RT, n (%)	13 (73)	7 (64)	6 (67)	26 (68)	.90#
Treatment before PsP, n (%)
Type of surgery					.98#
Biopsy	4 (22)	2 (18)	1 (11)	7 (18)
GTR	4 (22)	3 (27)	3 (33)	10 (26)
STR	10 (56)	6 (55)	5 (56)	21 (56)
Postoperative treatment					.53#
RT alone	2 (11)	0 (0)	0 (0)	2 (5)
RT + PCV	15 (83)	10 (91)	7 (78)	32 (84)
RT + Temozolomide	1 (6)	1 (9)	2 (22)	4 (11)
Clinical features at PsP, n (%)
New symptoms	1 (5)	1 (9)	0 (0)	2 (5)	.99#
KPS					.99#
≤70%	1 (6)	1 (9)	1 (11)	3 (8)
80%-100%	17 (94)	10 (91)	8 (89)	35 (92)
Radiological features of PsP, n (%)
New contrast-enhanced lesion	15 (83)	9 (82)	8 (89)	32 (84)	.99#
Increase of contrast-enhanced lesion	3 (7)	2 (18)	1 (11)	6 (16)	.99#
Pattern of contrast enhancement					.81#
Nodular	16 (89)	13 (93)	5 (83)	34 (89)
Ring-like	2 (11)	1 (7)	1 (17)	4 (11)
Location
Lobar	11 (61)	7 (64)	4 (44)	22 (58)	.49#
Deep-seated	4(22)	1 (9)	4 (44)	9 (24)
Both	3 (17)	3 (27)	1 (12)	7 (18)
Number of lesions					.13#
1	8 (44)	9 (82)	5 (83)	22 (58)
≥2	10 (56)	2 (18)	4 (17)	16 (42)
Size of the largest lesion					.10#
<1 cm	15 (83)	11 (100)	6 (67)	32 (84)
≥1 cm	3 (7)	0 (0)	3 (33)	6 (16)
No rCBV increase on MRI	12/13 (92)	6/6 (100)	6/7 (86)	24/26 (92)	1#
No 18F-DOPA PET hypermetabolism	5/5 (100)	1/1 (100)	1/2 (50)	7/8 (87)	.37#
Outcome
Regression, n (%)	12 (67)	7 (64)	5 (56)	24 (63)	.91#
Median time to regression, months (range)	7 (2-30)	3 (3-24)	4 (3-8)	6 (2-30)	.30°
OS, median months (95% CI)	121 (103-139)	81 (63-89)	NRa	121 (89-153)	<.001^
Subsequent progression, n (%)	1 (6)	2 (18)	2 (22)	5 (13)	.35#
5-year survival, %	100	91	89	94

Abbreviations: 95% CI, 95% confidence interval; A3, astrocytoma IDHmt grade 3; A4, astrocytoma IDHmt grade 4; CT, chemotherapy; GTR, gross total resection; KPS, Karnofsky performance status; NR, not reached; O3, oligodendroglioma IDHmt and 1p/19q codeleted grade 3; OS, overall survival; PCV, Procarbazine, CCNU, Vincristine; PET, positron emission tomography; PsP, pseudoprogression; rCBV, relative cerebral blood volume; RT, radiation therapy; STR, subtotal resection.

⁺Chi-squared test,

^#Fisher’s exact test or Fisher-Freeman-Halton exact test,

^{^}Log-rank test,

^°One-way ANOVA.

^aLast observation censored at 96 months (89% survival estimate).

Fig. 1.

Illustrative MRI of patients with a pseudoprogression (A) and outcome of patients with or without pseudoprogression (B) in the reference cohort. A: Examples of pseudoprogression. Upper panel: 42-year-old woman, grade 3 astrocytoma IDH1 R132H mutant, left: T1 post-contrast MRI performed 1 month after RT, middle: T1 post-contrast MRI performed 4 months after RT and 2 PCV cycles with perfusion MRI demonstrating no rCBV increase, right: T1 post-contrast MRI performed 10 months after RT and 6 PCV cycles, at last news the patient is still progression-free. Middle panel: 47-year-old man, grade 3 oligodendroglioma IDH1 R132H mutant and 1p19q codeleted, left: T1 post-contrast MRI performed 1 month after RT; middle: T1 post-contrast MRI performed 11 months after RT with perfusion MRI demonstrating no rCBV increase and no significant hypermetabolism on ¹⁸F-DOPA PET scan, right: T1 post-contrast MRI performed 43 months after RT, at last news the patient is still progression-free. Lower panel: 50-year-old man, grade 3 astrocytoma IDH1 R132H mutant, left: T1 post-contrast MRI performed 1 month after RT, middle: T1 post-contrast performed 15 months after RT and 6 temozolomide cycles with perfusion MRI demonstrating no rCBV increase, right: T1 post-contrast MRI performed 45 months after RT, at last news the patient is still progression-free. B: progression-free survival (PFS) and overall survival (OS) in patients with (green) and without (blue) pseudoprogression (PsP). In order to avoid a potential time bias (patients with a PsP have to live long enough to meet PsP definition), patients without PsP were excluded if they progressed (PFS analysis) or died (OS analysis) before the 95th percentile of the time to development of PsP (36 months). PFS and OS between patients with and without PsP were not significantly different. Abbreviations: OS, overall survival; PFS, progression-free survival; PsP, pseudoprogression.

The only characteristic significantly different according to the presence or absence of pseudoprogression was PCV CT (Table 3). Pseudoprogression was more frequent in patients who were treated with RT plus PCV than in those who were treated with RT alone or RT plus temozolomide (26% vs 7% vs 8%, respectively, P < .001). PFS and OS, corrected for a possible time bias, were similar in patients with or without pseudoprogression. However, there was a trend toward a higher probability of 5-year PFS in patients who presented a pseudoprogression (Table 3, Figure 1B).

Table 3.

Comparison of the Characteristics of Patient With or Without Pseudoprogression in the Reference Cohort

PsP	No PsP	P value
Number of patients with PsP, n/N (%)	38/200 (19)	162/200 (81)
Clinical characteristics at diagnosis
Sex ratio m/f, %	47/53	57/43	.29+
Median age, years (range)	42.5 (22-78)	41 (27/72)	.67°
KPS, median	100	100	.87°
Seizures, n (%)	30 (79)	115 (71)	.32+
Radiological characteristics at diagnosis, %
Frontal location	71	68	.76+
Contrast enhancement	68	67	.57+
Mass effect	59	61	.81+
Histo-molecular characteristics, n (%)
WHO 2016 integrated diagnosis			.82+
O3	18 (47)	81 (50)
A3	11 (29)	39 (24)
A4	9 (24)	42 (26)
IDH1 R132H mutation	35 (92)	144 (89)	.71+
Non-canonical IDH mutation	3 (8)	18 (11)	.77#
MGMT promoter methylation	14/17 (82)	45/62 (73)	.53#
TERT promoter mutation	15 (39)	76 (47)	.41+
CDKN2A homozygous deletion	2/36 (6)	14/158 (9)	.74#
9p loss	9/33 (27)	40/147 (27)	.99+
Treatment characteristics, n (%)
Type of surgery			.81+
Biopsy	7 (18)	30 (18)
GTR	10 (26)	35 (22)
STR	21 (56)	97 (60)
Postoperative treatment			<.001#
RT alone	2 (5)	25 (15)
RT + PCV	32 (84)	90 (56)
RT + Temozolomide	4 (11)	47 (29)
RT regimen			.11#
59.4 Gy in 33 fraction of 1.8 Gy	11 (29)	77 (48)
60 Gy in 30 fractions of 2 Gy	23 (61)	72 (44)
54 Gy in 30 fractions of 1.8 Gy	1 (2)	1 (1)
54 Gy in 27 fractions of 2 Gy	3 (8)	12 (7)
Outcome
OS, median months (95% CI)	121 (89-153)	NRa	.40^
With progression, n (%)	5 (13)	41 (25)	.09+
5-year PFS, %	87	76
5-year OS, %	94	86

Abbreviations: 95% CI, 95% confidence interval; A3, astrocytoma IDHmt grade 3; A4, astrocytoma IDHmt grade 4; CT, chemotherapy; GTR, gross total resection; KPS, Karnofsky performance status; m/f, male/female; NR, not reached; O3, oligodendroglioma IDHmt and 1p/19q codeleted grade 3; OS, overall survival; PCV, Procarbazine, CCNU, Vincristine; PFS, progression-free survival; PsP, pseudoprogression; RT, radiation therapy; STR, subtotal resection.

⁺Chi-squared test,

^#Fisher’s exact test or Fisher-Freeman-Halton exact test,

^{^}Log-rank test,

^°One-way ANOVA.

^aLast observation censored at 142 months (61% survival estimate).

Estimation of Pseudoprogression Misdiagnosis in the Reference Cohort

Since the distinction between pseudoprogression and true progression can be difficult, we wondered whether among patients diagnosed with a true progression, some patients could have had in fact a misdiagnosed pseudoprogression. We hypothesized that comparing PFS1 (before first progression) and PFS2 (after first progression) could identify these patients. To define the PFS2/PFS1 ratio suggestive of pseudoprogression misdiagnosis, we looked at PFS before and after pseudoprogression in the 38 patients with a definite pseudoprogression. We found that for all of the 30 patients for whom sufficient follow-up was available, the duration of PFS after pseudoprogression was at least twice as long as the time to pseudoprogression. We therefore hypothesized that a PFS2 at least twice as long as PFS1 (PFS2 ≥ 2×PFS1) associated with a prolonged survival (>2 years) could identify patients with a misdiagnosed pseudoprogression (Supplementary Figure S1).

We found that among the 41 patients considered as having a true progression and for whom sufficient follow-up was available, 8 patients had a PFS2 ≥ 2×PFS1 and 7 of these had a prolonged survival. These patients accounted for 17% of all patients considered to have a true progression and for 20% of the patients considered to have a true progression within the first 2 years after RT. The characteristics of these 7 cases strongly suggested a misdiagnosed pseudoprogression. For instance, despite an assumed early progression (median 10 months), 6 patients remained progression-free after a median of 58 months and the only patient who died did so from a cause other than tumor progression. In addition, this prolonged disease control was achieved after a second-line treatment that was less intense than the first-line treatment in 6 patients. Furthermore, among the 4 who underwent resurgery (at first progression in 3 patients and at second progression in 1 patient), important post-therapeutic changes were observed in the 3 patients. Finally, rCBV was not increased in 2 of the 3 patients who underwent a perfusion MRI (Table 4).

Table 4.

Characteristics of the 7 Patients in the Reference Cohort With a Likely Midiagnosed Pseudoprogression

Patient	Diagnosis	First-Line Treatment	Clinical Features at Progression	Radiologic Features at Progression	PFS 1 (mo)	Second-Line Treatment	Histological Analysis at Resurgery	PFS 2 (mo)	OS (mo)
1	A4	GTR RT + PCV (n = 6)	New deficit, KPS 80%	Ring-like CE, edema; no rCBV increase	23	TMZ (n = 14) + Bevacizumab (n = 28)	NA	58a	82
2	O3	Biopsy RT/TMZ (n = 30)	New deficit, KPS 80%	Not available	33	Bevacizumab	NA	84a	117
3	A4	Biopsy RT/TMZ (n = 6)	New deficit, KPS 80%	Cystic evolution, new CE	13	Surgery alone	Tumor	61a	74
4	A3	GTR RT + PCV (n = 3)	New deficit, KPS 80%	Ring-like CE, edema; rCBV increase	13	TMZ (n = 13) + Bevacizumab (n = 26)	NA	58a	72
5	O3	GTR RT + PCV (n = 2)	New deficit, KPS 70%	Not available	6	Surgery + TMZ (n = 18)	Tumor and post- therapeutic changes	44a	51
6	O3	Biopsy RT alone	Seizures, KPS 90%	Ring-like CE	29	Surgery + TMZ (n = 24)	Tumor and post- therapeutic changes	68b	98
7	O3	STR RT/TMZ (n = 9)	Seizures, KPS 90%	Cystic evolution, increase of CE, no rCBV increase	20	PCV (n = 6)	Tumor and post- therapeutic changesc	54	93

Abbreviations: A3, astrocytoma IDHmt grade 3; A4, astrocytoma IDHmt grade 4; CE, contrast enhancement; GTR, gross total resection; KPS, Karnofsky performance status; NA, not available; O3, oligodendroglioma IDHmt and 1p/19q codeleted grade 3; OS, overall survival; PCV, Procarbazine, CCNU, Vincristine; PFS, progression-free survival; rCBV; relative cerebral blood volume; RT, radiotherapy; TMZ; temozolomide.

^aNo second progression, alive at the end of follow-up.

^bNo second progression, death not in relation with tumor progression.

^cResurgery in this patient was performed at the time of second progression.

Pseudoprogression vs True Progression vs Misdiagnosed Pseudoprogression in the Reference Cohort

The characteristics of patients with a pseudoprogression differed from those with a true progression (Supplementary Table S1). Pseudoprogression occurred earlier after the end of RT than did true progression (10.5 vs 23 months, P < .001). Within the first 2 years after RT, the risk of developing pseudoprogression appeared to be greater than that of developing tumor progression (Supplementary Figure S3A). Compared to patients with a true progression, those with a pseudoprogression more frequently presented lesions <1 cm (respectively, 16% vs 97%, P < .001), less frequently presented new symptoms (respectively, 72% vs 5%, P < .001) and had a better median OS (51 vs 121 months, P < .001). Within the first 2 years after RT, patients diagnosed with a pseudoprogression had a much better outcome than those diagnosed with a true progression and an outcome similar to those who presented neither progression nor pseudoprogression (Supplementary Figure S3B). Regarding patients with a likely misdiagnosed pseudoprogression they were characterized by a time to event and a median OS similar to that of patients with a pseudoprogression (10 vs 10.5 months, P = .88, and 98 months vs 121 months, P = .48, respectively) but as patients with a true progression they more frequently presented with large and symptomatic lesions (illustrated in Figure 2A) compared to those diagnosed with a pseudoprogression (86% vs 16%, P < .001, and 100% vs 5%, P < .001, respectively).

Fig. 2.

Illustrative examples of MRI of patients with a likely misdiagnosed pseudoprogression in the reference cohort (A) and swimmer plot of patients suspected to have a misdiagnosed pseudoprogression in the control cohort (B). A: Examples of likely misdiagnosed pseudoprogression in the reference cohort. Upper panel: 38-year-old woman, grade 3 astrocytoma IDH1 R132H mutant, left: T1 post-contrast MRI performed at diagnosis, middle: T1 post-contrast MRI performed 6 months after RT and 3 PCV cycles, at this time the patient was treated with temozolomide and bevacizumab because of a suspicion of tumor progression, right: T1 post-contrast MRI performed 5.5 years later, the patient is still progression-free. Lower panel: 33-year-old women, grade 4 astrocytoma IDH1 R132H mutant, left: T1 post-contrast MRI performed at baseline, middle: T1 post-contrast MRI performed 10 months after RT and 6 temozolomide cycles, at this time the patient was treated by surgery alone because of a suspicion of tumor progression, right: T1 post-contrast MRI performed 6 years later, the patient is still progression-free. B: Swimmer plot of control cohort patients retrospectively suspected to have a misdiagnosed pseudoprogression illustrating the discordance between progression-free survival before (PFS1 orange) and after presumed progression (PFS2 gray). PFS3 is represented in yellow. Abbreviations: A3, astrocytoma IDHmt grade 3; A4, astrocytoma IDHmt grade 4; O3, oligodendroglioma IDHmt and 1p/19q codeleted grade 3; PFS, progression-free survival.

Estimation of Pseudoprogression Misdiagnosis in the Control Cohort

Since using PFS2 ≥ 2×PFS1 enabled us to identify patients with characteristics strongly suggestive of a misdiagnosed pseudoprogression in the reference cohort, we used the same criteria to estimate the rate of pseudoprogression misdiagnosis in the control cohort of 543 patients IDHmt HGG treated with RT (with or without CT). In this cohort, we found that 36 out of the 203 patients considered as having a true progression and for whom sufficient follow-up was available had a PFS2 ≥ 2×PFS1 and that among these, 35 had a prolonged survival (Figure 2B). These patients accounted for 17% of all patients considered to have a true progression and for 21% of the patients considered to have a true progression within the first 2 years after RT. Initial treatment in these patients consisted of RT plus CT in 25 patients (71%) and RT alone in 10 patients (29%). Second-line treatment consisted of CT alone (temozolomide n = 14, CCNU n = 3, PCV n = 2), resurgery alone (n = 6) or followed by CT (n = 4), bevacizumab alone (n = 2), simple follow-up (n = 3), and reirradiation (n = 1). Analysis of these 35 cases strongly suggested a misdiagnosed pseudoprogression. For instance, despite an early progression (median 12 months), 24 patients remained progression-free after a median duration of 59.5 months, and 28 were alive after a median follow-up of 59 months after the assumed progression. Also consistent with this hypothesis, as observed in the reference cohort, compared to patients with a PFS1 < 2×PFS2 (strongly suggestive of a true progression), those with a suspected misdiagnosed pseudoprogression had a shorter PFS1 (12 vs 19 months, respectively, P < .001) but a much better outcome (5-year survival of 87% vs 31%, P < .001).

Discussion

To the best of our knowledge, the present study is the first to specifically analyze pseudoprogression incidence and characteristics in IDHmt HGG patients. It suggests that in these patients, pseudoprogression is not only frequent but also frequently misdiagnosed as true progression, and that within the first 2 years after RT, the possibility of a pseudoprogression should be carefully considered.

Comparing the incidence of pseudoprogression across studies that focused on low-grade gliomas (LGG) and/or IDHmt gliomas is complicated as many studies included both LGG and HGG but also by the use of different definitions of pseudoprogression, in particular regarding the time window considered for the occurrence of pseudoprogression^10,25–35 (Supplementary Table S2). Herein, we used the same definition as van West et al in their study on LGG with the only difference that we accepted the use of corticosteroids.¹⁰ As these authors, we did not set up any maximal time limit for pseudoprogression which has the advantage of setting no a priori regarding the timing of pseudoprogression. Using this definition, the incidence of pseudoprogression in IDHmt HGG was in the same range as that reported in glioblastomas (2%-54%), LGG (12%-39%), brainstem gliomas (13%-37%), and circumscribed gliomas (40.6%).^{10,12,25,26,33–36} Pseudoprogression, however, occurred later in IDHmt HGG than in glioblastomas in which the typically reported timing is around 3-6 months.^1,5,9 The timing of pseudoprogression in IDHmt HGG was similar to the 12 months timing reported in LGG,¹⁰ which is consistent with the results of a study that found that treatment-related changes occurred later in IDHmt (typically >5 months) than in IDHwt gliomas (typically <5 months).¹³ It could also explain why some studies that focused only on treatment-related changes within the first 3-6 months after RT reported that pseudoprogression was less frequent in IDHmt and 1p/19q codeleted gliomas than in IDHwt glioblastomas.^27,28 Regarding pseudoprogression characteristics, these were very similar to that described in LGG with most patients displaying small and asymptomatic contrast-enhanced lesions and we observed no difference regarding IDHmt HGG subtypes.¹⁰

Another difficulty when studying pseudoprogression is that some cases can wrongly be misdiagnosed as true progression even by experienced teams. In addition, it has been shown that when patients undergo resurgery, the distinction based on histological analysis between pseudoprogression and true tumor progression is frequently difficult due to the presence of both post-therapeutic changes and residual tumor.⁶ van West et al reported that 3 out of the 31 patients (9.6%) in their cohort judged to have a true progression more likely had a misdiagnosed pseudoprogression because of their radiological characteristics and because of a PFS2 longer than PFS1.¹⁰ Indeed, it has been shown in IDH-mutant gliomas that PFS2 is usually shorter than PFS1.³⁷ Herein, based on the timing of pseudoprogression in our reference cohort, we used PFS2 ≥ 2×PFS1 as a criterion to identify patients with a potentially misdiagnosed pseudoprogression. Although it cannot be excluded that some patients had a true tumor progression followed by a prolonged response to second-line therapy, we found that in our reference cohort, patients who fulfilled this criterion had a history strongly suggestive of a misdiagnosed pseudoprogression. Interestingly, we found a similar rate of patients with a likely misdiagnosed pseudoprogression in our reference and in our control cohort. Analysis of the characteristics of patients with a likely misdiagnosed pseudoprogression suggested that these patients were misdiagnosed because they were symptomatic and had larger lesions than those usually observed in patients with a pseudoprogression. If one assumes that in our series, patients with a PFS2 ≥ 2×PFS1 had indeed a misdiagnosed pseudoprogression, this would mean that about 1 out of 5 patients who were judged to have true progression actually had a pseudoprogression. These findings support careful evaluation of IDHmt HGG patients suspected to have a first tumor progression within 2 years after RT. In our opinion, these patients should systematically undergo multimodal MRI, especially perfusion MRI, and if MRI results are uncertain they should undergo metabolic imaging such as ¹⁸F-DOPA PET scan. Following these investigations, if the possibility of a pseudoprogression is raised, asymptomatic patients with small lesions should be followed closely. In patients who are symptomatic and or have large lesions, resurgery should be considered; if this is not possible patients should be treated with steroids or bevacizumab. Perfusion MRI and ¹⁸F-DOPA PET scan are very helpful to distinguish pseudoprogression from true progression; however, patients with pseudoprogression can occasionally present with high rCBV and increased ¹⁸F-DOPA PET scan metabolism; this was the case herein for 8% and 13% of the patients, respectively. Regarding clinical trial design, the potential difficulty of pseudoprogression diagnosis in IDHmt HGG supports the implementation of randomized rather than single-arm studies for trials aiming to test new strategies at first recurrence and the requirement of multimodal imaging in the inclusion criteria.

In glioblastomas, retrospective studies suggested that pseudoprogression is more frequent in patients treated with RT and temozolomide and more frequent in those with a methylated MGMT promoter.^9,30,38–40 Herein, we found that the only characteristic significantly different between those with and without pseudoprogression was adjuvant PCV CT. This finding needs to be confirmed in an independent series but is consistent with a previous study reporting pseudoprogression after PCV CT.⁴¹ In LGG, pseudoprogression has been shown to be more frequent in patients treated with proton therapy than with IMRT.³³ Herein, there was a trend toward more frequent pseudoprogression in patients treated with 60 Gy in 30 fractions schedule than in those treated with 59.4 Gy in 33 fractions schedule.

The prognostic value of the occurrence of a pseudoprogression in low-grade and/or IDH-mutant gliomas remains controversial. Some studies reported a favorable impact on outcome^25,26 while others did not^10,29,30 (Supplementary Table S2). Herein, there was a trend toward a better PFS in patients who presented a pseudoprogression; however, the number of events in our study was low and a follow-up study will be needed to confirm this finding.

Another issue when studying pseudoprogression is the distinction between pseudoprogression and radiation necrosis. It is currently unclear in which way the pathophysiology and the timing of pseudoprogression are different. Herein, given the definition of pseudoprogression that we used, we qualified as pseudoprogression, post-therapeutic changes that occurred several years after RT and that could also have been qualified as radiation necrosis. Winter et al recently proposed to use the term pseudoprogression for post-therapeutic changes occurring <5 months and treatment-induced necrosis for those occurring >5 months after RT.^13,14 They reported differences regarding these two types of post-therapeutic changes especially that IDH-mutant gliomas were more likely to present with treatment-induced necrosis than with pseudoprogression, which is consistent with the timing of post-therapeutic changes in the present study. Voss et al also reported a series of glioma patients who presented stable or spontaneously disappearing new contrast enhancing spots (with a median onset of 30 months) and proposed to consider this post-therapeutic changes as a new pattern of late-onset pseudoprogression.⁴² Interestingly, this phenomenon was mostly observed in IDH-mutant gliomas (60%) which is also consistent with our findings.

Other than its retrospective design, the present study is limited by the fact that perfusion and metabolic imaging were only available in a small number of patients and were not centrally reviewed. Radiation therapy fields were not reviewed neither. The retrospective evaluation of pseudoprogression misdiagnosis based on PFS has the limitation of being speculative and of using a criterion (PFS2/PFS1) that remains to be validated. Despite these limitations, the present study, conducted in a large cohort of patients who were prospectively followed within a dedicated network, provides for the first time a detailed description of pseudoprogression in IDHmt HGG. It shows that pseudoprogression in these tumors is not only to be suspected within the first few months after RT completion and suggests that although pseudoprogression may be easily suspected when it consists of small asymptomatic lesions, its diagnosis may be much more challenging in symptomatic patients with large lesions. We believe that these findings are important for clinical practice but also for appropriate clinical trial design.

POLA Network Collaborators

Christine Desenclos (Hôpital Nord, CHU Amiens, Neurosurgery Department, Amiens, France), Henri Sevestre (Hôpital Nord, CHU Amiens, Pathology Department, Amiens, France), Philippe Menei (CHU Angers, Neurosurgery Department, Angers, France), Audrey Rousseau (CHU Angers, Pathology Department, Angers, France), Thierry Cruel (Mont-Blanc Pathology, Annecy, France), Stéphane Lopez (CH Annecy, Oncology Department, Annecy, France), Marcela Ionela Mihai (Hôpital Jean Minjoz, CHU Besançon, Pathology Department, Besançon, France), Antoine Petit (Hôpital Jean Minjoz, CHU Besançon, Neurosurgery Department, Besançon, France), Clovis Adam (Hôpital Bicêtre, Pathology Department, Le Kremlin-Bicêtre, France), Fabrice Parker (Hôpital Bicêtre, Neurosurgery Department, Le Kremlin-Bicêtre, France), Romuald Seizeur (Hôpital de la cavale blanche, CHU Brest, Neurosurgery Department, Brest, France), Isabelle Quintin-Roué (Hôpital de la cavale blanche, CHU Brest, Pathology Department, Brest, France), Sandrine Eimer (CHU de Bordeaux-GH Pellegrin, Pathology Department, Bordeaux, France), Damien Ricard (HIA Percy, Neurology Department, Clamart, France), Catherine Godfraind (CHU Clermont-Ferrand, Pathology Department, Clermont-Ferrand, France), Toufik Khallil (CHU Clermont-Ferrand, Neurosurgery Department, Clermont-Ferrand, France), Dominique Cazals-Hatem (Hôpital Beaujon, Pathology Department, Clichy, France), Thierry Faillot (Hôpital Beaujon, Neurosurgery Department, Clichy, France), Claude Gaultier (CH Colmar, Neurology Department, Colmar, France), Marie Christine Tortel (Hôpital Colmar, Pathology Department, Colmar, France), Ioana Carpiuc (Clinique des Cèdres, Medical Oncology Department, Cornebarrieu, France), Pomone Richard (Clinique des Cèdres, Pathology Department, Cornebarrieu, France), Marie-Hélène Aubriot-Lorton (CHU Dijon, Pathology Department, Dijon, France), François Ghiringhelli (Centre Georges-François Leclerc, Medical Oncology, Dijon, France), Claude-Alain Maurage (CHU de Lille, Pathology Department, Lille, France), Edouard Marcel Gueye (Hôpital Dupuytren, CHU de Limoges, Neurosurgery Department, Limoges, France), Francois Labrousse (Hôpital Dupuytren, CHU de Limoges, Pathology Department, Limoges, France), Luc Bauchet (CHU de Montpellier, Neurosurgery Department, Montpellier, France), Valérie Rigau (CHU de Montpellier, Pathology Department, Montpellier, France), Guillaume Gauchotte (CHU Nancy, Pathology Department, Nancy, France), Luc Taillandier (CHU Nancy, Neurology Department, Nancy, France), Mario Campone (Centre René Gauducheau, Medical Oncology Department, Saint-Herblain, France), Delphine Loussouarn (CHU Nantes, Pathology Department, Nantes, France), Véronique Bourg (CHU Nice, Neurology Department, Nice, France), Fanny Vandenbos-Burel (CHU Nice, Pathology Department, Nice, France), Claire Blechet (CHR Orléans, Pathology Department, Orléans, France), Mélanie Fesneau (CHR Orléans, Radiotherapy Department, Orléans, France), Franck Bielle (Hôpital Pitié-Salpêtriere, Neuropathology Department, Paris, France), Antoine Carpentier (Hôpital Saint-Louis, Neurology Department, Paris, France), Marc Polivka (Hôpital Lariboisière, AP-HP, Pathology Department, Paris, France), Serge Milin, (CHU Poitiers, Neurosurgery Department, Poitiers, France), Michel Wager (CHU Poitiers, Neurosurgery Department, Poitiers, France), Marie-Danièle Diebold (CHU Reims, Pathology Department, Reims, France), Danchristian Chiforeanu (CHU Rennes, Pathology Department, Rennes, France), Elodie Vauleon (Centre Eugène Marquis, Medical Oncology, Rennes, France), Annie Laquerriere (CHU Rouen, Pathology Department, Rennes, France), Fabien Forest (Hôpital Nord, CHU Saint-Étienne, Pathology Department, Saint-Priest en Jarez, France), Marie Janette Motso Fotso (Hôpital Nord, CHU Saint-Étienne, Neurosurgery Department, Saint-Priest en Jarez, France), Marie Andraud (CHU Saint-Pierre de la Réunion, Pathology Department, Saint-Pierre de la Réunion, France), Gwenaelle Runavot (CHU Saint-Pierre de la Réunion, Neurology Department, Saint-Pierre de la Réunion, France), Benoit Lhermitte (CHU Strasbourg, Pathology Department, Strasbourg, France), Georges Noel (Centre Paul Strauss, Radiotherapy Department, Strasbourg, France), Michèle Bernier (Hôpital Foch, Neurology Department, Suresnes, France), Chiara Villa (Hôpital Foch, Pathology Department, Suresnes, France), Cécilia Rousselot-Denis (CHU Bretonneau, Pathology Department Tours, France), Ilyess Zemmoura (CHU Bretonneau, Neurosurgery Department Tours, France), Nicolas Desse (HIA Sainte-Anne, Neurosurgery Department, Toulon, France), Emmanuelle Uro-Coste (Toulouse University Hospital, Pathology Department, Toulouse, France).

Funding

None.

Conflict of interest statement. None.

Authorship statement. Data collection and analysis: A.S., F.D., and C.D. Interpretation of analysis: all authors. Drafting of initial manuscript: A.S., F.D., and C.D. Reviewing, editing, and approval for final manuscript: all authors.

Ethics statement. All patients provided written informed consent for clinical data collection and genetic analysis according to national and POLA network policies.