Introduction

The management of low grade diffuse gliomas (WHO grade II, LGGs) is one of the most controversial areas of neuro-oncology. These tumors frequently present with readily-controlled seizures in patients who are often young, and have no fixed neurological deficit. They behave indolently for an unpredictable period, which may be many years, before undergoing malignant transformation into high grade lesions which are rapidly fatal. Prospective evidence for survival advantage from early intervention is weak, and although total or subtotal resection may confer benefit, this is frequently not possible due to eloquent anatomical location and the infiltrative nature of the tumor. The potential benefit of early aggressive treatment must therefore be balanced against the morbidity associated with investigation and treatment in patients who are frequently clinically well, and may not develop significant problems for several years. In many institutions, patients with LGG are therefore monitored on a “watch and wait” policy until there are signs of malignant transformation, at which point radiotherapy and/or surgery is commenced. Conventional techniques for assessing gliomas include contrast enhanced MRI and histopathological assessment based on stereotactic biopsy or bulk resection, although limitations with these methods are well recognised in the literature (Knopp et al., 1999; Coons et al., 1997). The development of more reliable methods for characterisation and risk stratification of LGG is important, to aid clinical decision making.

Perfusion Magnetic Resonance (MR) imaging techniques allow calculation of cerebral blood volume (CBV), which is a measure of localised cerebral blood volume within tissue. Changes in the CBV of gliomas can be explained by tumor pathophysiology; neoangiogenesis is associated with an increase in vessel density, tortuosity and permeability, and are a hallmark of aggressive elements in histopathological analysis of gliomas. CBV therefore provides a non-invasive quantitative imaging biomarker that reflects angiogenesis in gliomas, and hence can aid characterisation of gliomas. In particular, it may be helpful in grading gliomas and identifying malignant transformation at an earlier stage.

Acquisition of rCBV

Dynamic susceptibility contrast enhanced MRI (DSC-MRI) is the most common perfusion method used in clinical practice, and the technique on which most published relative cerebral blood volume (rCBV) data in gliomas has been based. The method involves acquiring time-resolved images during the transit of an exogenous gadolinium chelate (Gd) contrast agent through the brain. Local magnetic field inhomogeneities caused by the Gd leads to a transient reduction in the transverse relaxation time of water protons in adjacent tissues, and a consequent reduction in tissue signal intensity (T2* effect). The concentration of agent is proportional to the change in relaxation rate, and analysis of the kinetics of signal intensity change allows perfusion parameters to be derived. In order to sample the signal profile from the first pass of contrast agent, the compound is injected intravenously as a rapid bolus, and rapid echo planar imaging (EPI) based spin echo or gradient echo sequences allow the whole brain to be imaged with sufficient time-resolution.

CBV is proportional to the area under the deconvolved time intensity curve, assuming there has been no recirculation or contrast leakage. Because the blood-brain barrier of contrast-enhancing tumors is leaky, contrast leakage needs to be accounted for when CBV values are calculated. A significant correlation between glioma grade and CBV values corrected for contrast extravasation has been previously demonstrated, but no correlation for uncorrected values was found (Boxerman et al., 2006). CBV is the proportion of the volume of interest which is comprised of vessels. It is measured as millilitres of blood per grams of tissue. Because of difficulties in reproducible absolute quantification of CBV, relative CBV is usually expressed as a ratio of CBV within the tissue of interest against normal contralateral white matter; this facilitates stable longitudinal perfusion measurement and inter-subject comparisons. Comparison of quantitative data across institutions and different MRI manufacturer platforms, however, remain a challenge.

Perfusion parameters can also be derived from the first pass phase of dynamic contrast enhanced MRI, which is based on T1-dependent signal abnormality, and also yields permeability data. These methods have been used to study gliomas, but are less widely validated. Arterial Spin Labelling is an alternative perfusion technique which uses “magnetically-labelled” water as an endogenous tracer. Water is labelled by inversion or saturation pulses and imaging is performed downstream from the site of labelled water to allow flow to the region of interest. The technique offers the advantage of being completely non-invasive and allows absolute quantification of blood flow, but its clinical application has been limited by long acquisition times and requirements for complex off-line processing. Technical advances with the current generation of commercial 3T MRI systems and more user-friendly processing software are likely to allow wider clinical application of ASL in future.

Comparison of rCBV and Conventional MRI for Glioma Evaluation

Several studies have demonstrated that rCBV can give useful information about tumor grade and patient prognosis (Table 41.1). Accurate characterisation of gliomas is important to allow appropriate risk stratification, which will inform difficult decisions on potentially invasive investigation and treatment. Conventional assessment of gliomas includes evaluation with contrast enhanced MRI sequences and histopathological evaluation following biopsy or tumor resection. Limitations of conventional methods of assessment are well recognised in the literature (Knopp et al., 1999; Coons et al., 1997). Relative CBV gives supplementary information on tumor physiology which complements that from conventional contrast enhanced MRI and can potentially detect signs of transformation earlier than conventional methods (Figs. 41.1, 41.2, and 41.3). Relative CBV is related to perfusion, and is felt to reflect angiogenesis, which is an important marker of tumor aggressiveness in the pathological grading of gliomas. Relative CBV can therefore differentiate biologically aggressive tumors from more indolent tumors and guide the most appropriate management. Contrast enhancement is often used as an indicator of progression on conventional MRI studies, but this can lead to an inaccurate assessment of the tumor (Knopp et al., 1999). Enhancement represents loss of the local blood-brain barrier but does not necessarily give a good indication of vascularity. Lack of enhancement does not always equate to a low grade tumor (Fig. 41.3) and variability in enhancement patterns between subgroups of low grade and high grade gliomas is well recognised (White et al., 2005; Ginsberg et al., 1998). Other characteristics on conventional MRI such as mass effect, oedema, necrosis and haemorrhage also lack accuracy in differentiating gliomas by grade (Aronen et al., 1994; Law et al., 2003; Rollin et al., 2006).

Fig. 41.1
figure 41_1_215673_1_En

Grade 2 astrocytoma. Left: T1-weighted contrast-enhanced (T1CE) study demonstrates minimal contrast enhancement. Right: The rCBV map demonstrates low rCBV values

Full size image
Fig. 41.2
figure 41_2_215673_1_En

Glioblastoma multiforme. Left: T1CE study demonstrates marked contrast enhancement around the tumour. Right: The rCBV map demonstrates elevated rCBV values

Full size image
Fig. 41.3
figure 41_3_215673_1_En

Glioblastoma multiforme. The rCBV map (right) shows elevated rCBV values before any significant contrast enhancement is demonstrated on the T1CE study (left)

Full size image
Table 41.1 rCBV values suggested by different authors for characterising gliomas
Full size table

Differentiating Gliomas Using rCBV

Many groups have attempted to determine rCBV measurements that will discriminate LGGs from high grade gliomas (HGGs). Table 41.1 summarises the results from 15 studies which analysed tumors using rCBV. The range of mean maximal rCBV values for LGGs is 1.11–2.14. The range of mean maximal rCBV values for HGGs is 2.47–6.10. A wide range of threshold values were suggested to discriminate between low grade and high grade (1.5–3.8).

Challenges of Glioma Evaluation with rCBV

For use in routine clinical practice, consensus on a precise rCBV value is required to distinguish between low grade and high grade gliomas. However, Table 41.1 demonstrates a range of values have been suggested in the literature, which can be attributed to a number of factors.

Technical differences in acquisition and analytical methods make comparison and synthesis of data from these studies difficult. A study found CBV values were higher using Gradient Echo – Echo Planar imaging (GE EPI) compared with Spin Echo – Echo Planar imaging (SE EPI) (Sugahara et al., 2001). GE EPI has been preferred to SE EPI by some groups because it is more sensitive to the contribution of vessels of a range of sizes whilst SE derived imaging has been reported as being more sensitive for capillary sized vessels (Boxerman et al., 2006; Sugahara et al., 2001). High grade tumors contain larger calibre vessels and studies have suggested GE rCBV values show significant correlation with tumor grade whilst SE rCBV values do not (Boxerman et al., 2006). The technique for selecting regions of interest (ROIs) may also affect values (Lev et al., 2004). A relatively small “hot spot” surrounded by lower rCBVs will be perceived as a lower value if a larger ROI is used than a smaller ROI. Inclusion or exclusion of visible intratumoural vessels also affects the magnitude of calculated rCBV and hence thresholds for discriminating between tumor groups (Caseiras et al., 2008).

Variability in the subgroups of gliomas that were studied may also contribute to the differences. A difference in tumor architecture between astrocytomas and oligodendrogliomas is likely to cause an intrinsic difference in the rCBV of tumors of comparable grade. Maximum rCBV of low grade oligodendrogliomas have demonstrated higher values than low grade astrocytomas, which is thought to be because the former have a higher density of capillaries than the latter (Lev et al., 2004; Cha et al., 2005); moreover, deletion of chromosome 1p has also been associated with elevated rCBV values (Jenkinson et al., 2006). Relative CBV is a sensitive discriminant for 1p 19q deletion status, the genetic hallmark of oligodendroglioma features, in LGGs. This is, in effect, a confounder of reliable tumor grading in a lesion of unknown subtype; it may not be possible to distinguish between aggressive elements within a low grade astrocytoma and a stable low grade oligodendroglioma. Pilocytic astrocytomas have been excluded from some studies because they can demonstrate high rCBV despite their classification as grade 1 gliomas (Arvinda et al., 2009); fortunately they are frequently morphologically distinctive on structural imaging, although may occasionally mimic high grade diffuse gliomas.

Differences in the management of patient cohorts between studies will also affect rCBV. Treatment with steroids can decrease tumor permeability and blood volume measurements. Changes in permeability have been described at one hour after steroid administration, with a 15% decrease in the rCBV of peritumoral tissue (Ostergaard et al., 1999).

The level at which to set a threshold value will depend on what is felt to be the optimum balance between acceptable false positives and false negative rates. A high false positive rate will lead to potentially unnecessary investigation and treatment with associated risks. A high false negative rate will lead to inappropriate management of patients at high risk. The judgement of the most appropriate balance is likely to vary between authors.

The choice of reference point used to determine threshold values also varies between studies. Arguments have been made for the use of hard clinical endpoints rather than histology to determine threshold values (Bisdas et al., 2009). A clinical endpoint may not necessarily correlate well with grade and there are concerns over the validity of the use of histology as a “gold standard”. Histological grading of gliomas can be difficult, particularly when a tumor demonstrates an atypical pattern; suboptimal reproducibility and interobserver variation are recognised challenges (Coons et al., 1997). There is also the risk of sampling a misrepresentative area of tumor, which will result in misleading histopathological analysis in heterogeneous lesions; this typically results in an underestimate of grade or aggressiveness.

As with other methods, characterising tumor heterogeneity with perfusion presents challenges. A commonly used technique is to analyse the brightest discrete region, although such an area may not always be apparent, and the decision on the most appropriate region of interest for analysis may vary between observers. Some groups have overcome this by analysing the whole tumor, although these methods are more labor-intensive and time consuming.

The Role of rCBV Within a Multimodalilty Approach to Glioma Evaluation

Opinions on the performance of rCBV compared to other advanced imaging techniques for the evaluation of LGGs vary, but there is general consensus that the future role of rCBV is likely to be within the framework of a multimodal approach which incorporates both conventional MRI and other advanced MRI techniques. Such an approach is advantageous because it allows interrogation of different aspects of tumor biology, growth rate and tumor metabolism, in addition to angiogenesis. Amalgamation of this information is likely to give a more accurate estimation of risk of transformation than the individual techniques, which have their own intrinsic weaknesses.

Some comparisons between rCBV measurements, diffusion imaging, MR Spectroscopy (MRS) and conventional MR imaging have found rCBV to be the best performing parameter to distinguish low grade from high grade gliomas (Law et al., 2003; Zonari et al., 2007), and for distinguishing gliomas for other brain tumors (Weber et al., 2006). MRS was found to be better than rCBV for regions near the cortex because the rCBV of grey matter is less distinct from tumor values than white matter (Henry et al., 2000), although coregistration of rCBV maps with structural images can help to distinguish tumor signal from that of adjacent cortex. The evidence for rCBV as a marker of tumor behaviour and grade is best established in diffuse gliomas of astrocytic lineage. A borderline significant elevation in choline/creatine ratio on MRS was demonstrated 12 months before rCBV changes (p = 0.06) in oligodendrogliomas (Hlaihel et al., 2010); other techniques may be of particular use in characterising this tumor subtype, in which rCBV is known to correlate less reliably with malignancy. A study of a heterogeneous group of paediatric tumors has also reported changes in choline adjacent to enhancing tumor bed without corresponding changes in rCBV (Tzika et al., 2002). A comparison of tumor volume growth over a 6 month interval with a number of other parameters, including rCBV but not MRS measurements, found tumor growth at 6 months was the best overall predictor for time to transformation in LGGs (Caseiras et al., 2009), although comparative risk stratification over short and long time periods was not investigated.

Some groups have attempted to overcome the limitations of individual advanced MR techniques by developing algorithms which use a multimodality approach. Groups have analysed sensitivity, specificity and predictive value of tumor grading following addition of rCBV and MRS findings to conventional MRI, and concluded that accuracy increases (Zonari et al., 2007; Law et al., 2003; Arvinda et al., 2009). A multimodal algorithm has been suggested, which uses different techniques in a systematic manner to narrow the differential for an unknown intra-axial lesion (Al-Okaili et al., 2006).

Conclusion

Relative CBV measurements offer a validated, non-invasive method of characterising and monitoring LGGs. Standardisation of techniques for acquisition of rCBV, consensus on the optimal threshold value to discriminate gliomas and appreciation of the variability in rCBV between subgroups of gliomas will facilitate its use in routine clinical practice. Current evidence suggests variable performance, with rCBV demonstrating strengths in some areas but weaknesses in others when comparison is made with other advanced MRI techniques. The future role of rCBV is likely to be as part of a multimodal approach to glioma characterisation and monitoring.