Diffusion-Weighted Imaging Parameters to Track Success of Pyogenic Brain Abscess Therapy

Sandy Chen-Yu Chen; Hsiao-Wen Chung

The fundamentals of successful treatment of any disease include early diagnosis, timely treatment, and vigilant monitoring of response to treatment. To accomplish this sometimes-difficult process, it is crucial to employ the means that provide important disease markers, including laboratory parameters and imaging characteristics, in addition to clinical symptoms and signs. A brain abscess, for instance, demonstrates the value of improved medical imaging. Before cross-sectional imaging, the diagnosis of a brain abscess depended heavily on clinical history (usually a history of otitis media) and the presenting symptoms (nonspecific headache, fever, or both). Mortality from abscess decreased dramatically to nearly zero in the post-CT era. One may ask how CT has influenced the treatment of brain abscesses. To answer this question, a thorough understanding of the evolution of a brain abscess is essential.

The formation of a brain abscess follows a typical evolution that can be divided into four contiguous stages: early cerebritis (days 1–3), late cerebritis (days 4–9), early capsule (days 10–14), and late capsule (day 14 or later). The evolution involves the inflammatory responses of the brain to restrain microorganisms from spreading by eliciting local inflammatory cell infiltration and edema, and later, by the formation of a distinct collagenous capsule. A typically mature abscess consists of a thick collagenous capsule and an acidic medium inside the abscess that hinders effective treatment with intravenous administration of antimicrobials.

Conservative treatment of a brain abscess is most effective at the early stage of cerebritis, and surgical intervention using image-guided burr hole aspiration or excision after craniotomy is usually chosen for the disease management at the capsular stage. Although other factors such as size, location, number (multiple vs single), and type (multiloculated vs uniloculated or pyogenic vs nonpyogenic) may change the treatment of choice for brain abscesses, the general concept of the management is well understood: the selection of treatment methods depends on the evolution of the disease. Hence, it is desirable that a reliable imaging tool is available to reveal the stage of the abscess and to follow progression after treatment. CT and conventional MR imaging have served this purpose well for decades. The evolution of a brain abscess can be subdivided into three distinct radiological stages: early cerebritis shows local edema without contrast enhancement; late cerebritis, an ill-defined capsule of enhancing rim; and capsule, a distinctly enhancing capsule. However, because other diseases share similar imaging features, improved imaging strategies are desirable.

Until recently, diffusion-weighted (DW) imaging has proved useful in the diagnosis of pyogenic brain abscesses by showing a low apparent diffusion coefficient (ADC) in the abscess cavity. Although the cause of ADC decrease may be controversial, there is no doubt that this imaging sequence is helpful in differentiating brain abscesses from necrotic brain tumors and in establishing the diagnosis of the disease at the capsular stage when pus accumulates in the center of the lesion. More recently, a limited number of case reports have shown that DW imaging can depict the disease even earlier at cerebritis. If this could be further validated by experimental or larger-cohort studies, more patients could be conservatively treated with good outcomes.

In this issue of the AJNR, Cartes-Zumelzu and colleagues extend the use of DW imaging in seven patients undergoing therapy for confirmed brain abscesses. They found the reaccumulation of pus in the abscess cavity can be reliably depicted by DW imaging by showing decreased ADC 1 week after the initial normalization or increase of ADC following aspirations. The important implications of the study are twofold: first, DW imaging increases the specificity for the detection of pus reaccumulations compared with that of other MR imaging sequences and laboratory parameters for inflammation; second, it changes the clinical management of the disease and possibly shortens the course of therapy, thanks to the highly accurate CT-guided aspiration technique that can be easily repeated. On the other hand, it has long been known that contrast-enhanced CT and conventional MR imaging are not specific, although highly sensitive as compared with other clinical indices, for the detection of pus reaccumulation. In other words, they do not show changes of abscess contents over time after treatment. The typically rimlike enhancement of a mature abscess may not show improvement on CT scans or MR images for up to 5 weeks or longer after antibiotic treatment or aspiration. Other imaging parameters such as decrease of brain edema, abscess size, and improvement of inflammatory laboratory indices therefore have to be taken into account for possible alternative therapy if the patient does not improve. In Thurnher’s study, the reaccumulated pus appeared hyperintense on follow-up DW images (with low ADC values), but the contrast-enhanced T1-weighted MR images remained unchanged, as shown in pretreatment study. This highlights the merit of DW imaging in the evaluation of the brain abscess during therapy.

Although DW imaging appears promising in the follow-up of evolution of brain abscesses during treatment, questions surrounding the sensitivity and specificity of DW imaging in the detection of pus accumulation remain to be answered. It has been reported that small abscesses could be falsely negative at diffusion-weighted imaging. This has been attributed to the different intrinsic contents the small abscess has as compared with the larger abscesses. Abscesses caused by nonpyogenic pathogens such as fungus or parasites (eg, toxoplasmosis) may show increased ADC in the abscess cavity. Another important question is whether the hyperintense diffusion-weighted signals actually represent trauma-related accumulation after surgical aspiration. To exclude the possibility of blood accumulation in the previous abscess cavity, a T2*-weighted gradient echo imaging technique may be useful to make the differentiation.

Antibiotic treatment of brain abscesses generally takes from weeks to months because of the extended time needed for brain tissue to repair and close the abscess space. Therefore, it is worthwhile knowing the time course of DW imaging signal intensity change with respect to the dosage and duration of antibiotic treatment. This may help in establishing the medical treatment strategy and the DW imaging follow-up protocol. In Thurnher’s series, one patient (patient 4) was treated with only antibiotics. The DW imaging signal intensity returned to normal 1 week following treatment. It is not known whether patient 4 underwent a shorter-than-average treatment regimen or whether the patient received a smaller-than-normal antibiotic dose. On the other hand, the treatment decision for patients 3 and 6 are worth mentioning. Patient 3 showed relatively few changes in ADC from 0.53 to 0.81 × 10 ⁻³ mm2/s on the first follow-up image; thus, a surgical drain was placed. However, similarly stable ADCs for both abscesses in patient 6 from initial to follow-up imaging (from 0.36 to 0.60 and 0.41 to 0.51 × 10 ⁻³ mm2/s 1 week after drainage) led to a continuous follow-up without aggressive intervention. Such an inconsistency prompts the need for a large-cohort, prospective study in which, ideally, a threshold for ADC evolution considered with clinical status and laboratory data would indicate further alternative therapy.

In addition to the issues raised, what would be pertinent about DW imaging features of the response to brain abscess therapy would be the parameters from which an appropriate treatment course could be suggested. In particular, is it the absolute ADC value or its relative temporal change that reflects the reaccumulation of pus? This question could be addressed if a proper choice of the b value existed or even if high-b-value DW imaging was used. Likewise, is it the ADC value or the DW imaging signal intensity that is clinically important for abscesses? Thurnher et al’s data from the first follow-up images obtained in patients 3 and 6 showed that DW imaging signal intensity correlated with ADC. Although it is unclear whether ADC or DW imaging findings were considered by Thurnher et al to be most important threshfold for determining the success of surgical drainage, these results certainly emphasize the importance of examining both parameters rather than focusing on either one alone. DW imaging or ADC by themselves might not be specific enough to measure therapeutic response, particularly since it is unclear whether artifactual ADC values might arise from echo planar susceptibility effects when the abscesses are located near the skull base. All these unsolved issues remain to be answered by further investigation.