The old belief that DLGG may alternate indolent and growing periods is still alive, although not supported by any relevant studies. A probable explanation is that a minor increase of tumor diameter is difficult to identify just by “eyeball” qualitative comparison of two consecutive MRI (see Fig. 16.2). Tumor diameter can be measured by different techniques, either with linear measurements of one, two or three diameters, or it can be deduced from a full 3D volumetric segmentation. It has been clearly evidenced that estimating tumor size by segmenting (manually or semi-automatically) each axial slice on a computer reduces greatly the intra- and inter-reader variability [41]. Indeed, as stated in [42]: “DLGG are often irregular in shape and grow anisotropically, resulting in poor reproducibility of area or volume estimation based on linear measurements.” Moreover, a recent study reported that 3D volumetric segmentation are much less sensitive to head position in the MRI than 2D diameters [43]. The increasing availability of softwares allowing to perform segmentations on DICOM images (be they on dedicated stations in neuroradiology and radiotherapy department, in neuronavigations systems or even on PC—for e.g., Osirix, ImageJ,…) renders any other technique based on one, two or three diameters measurements oldfashioned [43]. Segmented volumes are then converted in diameters, by computing a volumetry-based diameter (d _V ):

Finally, the growth curve of this equivalent diameter can be plotted over time and the growth rate of the radiological diameter d_V, can be estimated for each patient from a simple linear regression, giving the Velocity of Diametric Expansion (VDE).

This methodology allowed to prove that DLGG are not radiologically stable. Mandonnet et al. [15] first showed quantitatively the spontaneous radiological growth of DLGG on successive MRIs in a subset of 27 histologically proven DLGG that were followed before oncological treatment. The average VDE was close to 4 mm/year, with a minimum at 2 mm/year. These initial results were confirmed by the same group on a larger series of 143 histologically proven DLGG, that ranged individual VDE from 1 to 36 mm/year [16], with a median VDE at 4.4 mm/year. In the series of Ricard et al. [17], the 39 patients with an available follow-up before chemotherapy also had a median VDE of 4.4 mm/year. Several independent groups have recently confirmed these results in series of patients harboring histologically proven supratentorial DLGG, as summarized in Table 16.1:

Finally, as already stated, two studies have evidenced a continuous growth of diameter in series of incidentally discovered DLGG:

Thus, these quantitative studies never reported a case of an indolent untreated DLGG with a stable tumor volume and a null VDE or a case of an untreated DLGG alternating indolent and growing periods. In summary, DLGG present a systematic, spontaneous and continuous radiological growth (although sometimes as slow as 1 mm/year), before any transformation into a higher grade of malignancy.

Several studies have shown that VDE greater than 8 mm/year on initial follow-up is highly suggestive of an imminent malignant transformation [16, 18, 24, 44]. However, for DLGG with an initial growth rate at 4 mm/year, it has never been shown whether malignant progression is preceded by an increase of the growth rate or not. This point will be further discussed in the chapter on biomathematical modeling. Advances in modern imaging methods,² including perfusion MRI and spectroscopic MRI, when performed at initial diagnosis, can yield valuable data regarding the anaplastic risk of an individual tumor [45], as detailed in the previous chapters (see Chap. 14). The aim of this paragraph is to put these results in a dynamic perspective and to emphasize what additional information can be gained from a longitudinal application of these techniques.

A striking feature of DLGG patients is that they do not have any focal neurological deficit. This means that brain networks plasticity can cope with lesions growing up to 4 mm/year, without any major consequence on motor or language function. However, studies with extensive assessment of cognitive status have evidenced that healthy controls scored better than DLGG patients [62]. This holds true even in patients with incidental DLGG [63]. Hence, it can be suspected that the decline in cognitive status of DLGG patients is a continuous process during the silent and low-grade period, as for the MRI evolution. However, very few studies have focused on the longitudinal cognitive follow-up of “wait and see” cohorts. Indeed, only one longitudinal study has been published, showing a worsening in non verbal delay recall scores after a one year “wait and see” follow-up [64]. Of course, at malignant transformation, the brain plasticity capabilities are overwhelmed by the fast growing tumor (VDE greater than 8 mm/year). At that time, focal neurological deficits, together with an increased seizures frequency, are commonly observed. All together, one can assume that the tumoral dynamics (as measured by the VDE) should be correlated with the dynamics of cognitive deterioration: the higher the tumor growth rate, the faster the cognitive deterioration. This hypothesis would deserve more clinical studies.

First of all, the different DLGG histological subtypes (oligodendroglioma, astrocytoma, mixed glioma) do not exhibit significant differences regarding the radiological tumor growth rates, as previously demonstrated in several studies [16–18, 65].

Three studies have investigated the link between DLGG genetic subtypes. In the first one [17], it was shown that DLGG with 1p-19q codeletion grew significantly slower than DLGG without (median VDE at 3.4 mm/year), and that DLGG with immunohistochemical overexpression of p53 grew significantly faster than DLGG without (median VDE at 4.2 mm/year). The second study [22] confirmed that growth rates of DLGG are lower when 1p-19q codeletion is present, whereas IDH status does not influence growth rates, a result which has been confirmed recently by a third study on a larger cohort [24]. This suggests that the favorable outcome of 1p-19q codeleted patients might be in part related to a tumor inherently more indolent, and that on the contrary, the good prognostic value of IDH mutation could result from a better efficacy of treatments or a lesser risk of malignant progression.

Only one study investigated quantitatively the effects of pregnancy on the radiological growth rates of DLGG [66, 67]. The results showed that DLGG accelerated significantly their radiological growth rates during pregnancy, above levels detected either before pregnancy or after delivery in 75% of cases. These changes in tumor growth were associated with an increase in seizure frequency in 40% of cases and radiological and clinical changes during pregnancy motivated further oncological treatment after delivery in 25% of cases. These results also underline that young women with DLGG should be informed of the oncogenic role of pregnancy (see Chap. 30).

The first study focusing on the prognostic significance of spontaneous MRI growth rates on overall survival was conducted on a retrospective series of 143 DLGG with measurements of the evolution of the diameter over time [16]. Overall survival was significantly higher in the low growth rates subgroup (median survival of 15 years for a VDE lower than 8 mm/year) than in the high growth rates subgroup (median survival of 5.6 years for a VDE at 8 mm/year or more).

The prognostic significance of spontaneous MRI growth rates on predicting progression into a higher grade of malignancy was addressed in two recent prospective studies. Brasil Caseiras et al. [18] proved in a series of 34 patients that “tumor growth within 6 months was better than baseline volumes, relative cerebral blood volume, or apparent diffusion coefficient in predicting time to malignant transformation in untreated DLGGs and was independent of other parameters”. They found a threshold of 6.21 mL of growth within 6 months, with a mean time of progression of 3.91 years versus 1.82 years. However, the prognostic significance of the evolution of the tumor volume (and not diameter) over time may be a combination of two independent other prognostic factors, the initial tumor volume and the tumor growth rate [68]. Or stated differently, the same amount of volumic increase may correspond to a large tumor with a low growth rate (of diameter) or a small tumor with a high growth rate (of diameter) [68]. Thus, VDE, obtained by the evolution over time of the diameter (deduced from the total segmented volume V) appears as a more reliable parameter than the evolution of the tumor volume to assess selectively the prognostic significance of radiological tumor growth rates.

Hlaihel et al. confirmed these results and demonstrated that an elevated VDE higher than 3 mm/year was correlated with a greater risk of progression into a higher grade of malignancy, with an average VDE at 7.87 mm/year in transformers group versus an average VDE at 2.14 mm/year in non transformers group [19]. The VDE threshold at about 8 mm/year is thus as a strong predictor of both malignant progression-free and overall survivals.

As DLGG are slow growing progressive tumors, with considerable variability in patient characteristics and therapeutic modalities, the accurate evaluation of the different oncological treatments efficacy constitutes a clinical challenge. Along with clinical response—particularly on seizure frequency—we will show that the quantitative assessment of the individual VDE by diameter evolution over time on consecutive MRI is a useful adjunct in the follow-up armamentarium.