記住我
ORIGINAL ARTICLE
Year : 2022 | Volume
: 8
| Issue : 4 | Page : 207-214
Effectiveness of pharmacologic interventions for prevention of cerebral hyperperfusion syndrome following bypass surgery
Georgios P Skandalakis1, Aristotelis Kalyvas2, Evgenia Lani3, Spyridon Komaitis3, Danai Manolakou4, Despoina Chatzopoulou5, Nikos Pantazis6, Georgios A Zenonos7, Constantinos G Hadjipanayis8, George Stranjalis3, Christos Koutsarnakis3
1 Athens Microneurosurgery Laboratory, Department of Neurosurgery, University of Athens, Athens, Greece; Department of Neurosurgery, Icahn School of Medicine, Mount Sinai Health System, New York, NY, USA
2 Athens Microneurosurgery Laboratory, Department of Neurosurgery, University of Athens, Athens, Greece; Department of Neurosurgery, Toronto Western Hospital, University Health Network, Toronto, Ontario, Canada, Switzerland
3 Athens Microneurosurgery Laboratory, Department of Neurosurgery, University of Athens; Department of Neurosurgery, National and Kapodistrian University of Athens School of Medicine, Athens, Greece
4 Department of Neurology, Hôpitaux Universitaires de Genève, Geneva University Hospitals, Geneva, Switzerland
5 Athens Microneurosurgery Laboratory, Department of Neurosurgery, University of Athens, Athens, Greece
6 Department of Hygiene, Epidemiology and Medical Statistics, School of Medicine, National and Kapodistrian University of Athens, Athens, Greece
7 Department of Neurosurgery, University of Pittsburgh Medical Center, Pittsburgh, PA, USA
8 Department of Neurosurgery, Icahn School of Medicine, Mount Sinai Health System, New York, NY; Department of Neurosurgery, University of Pittsburgh Medical Center, Pittsburgh, PA; Department of Neurosurgery, Icahn School of Medicine Mount Sinai Beth Israel, Mount Sinai Health System, New York, NY, USA
Correspondence Address:
Georgios P Skandalakis
Department of Neurosurgery, University of Athens School of Medicine, Athens, Greece.
Source of Support: None, Conflict of Interest: None
CheckDOI: 10.4103/bc.bc_43_22
BACKGROUND: Cerebral hyperperfusion syndrome (CHS) following bypass surgery is a major cause of neurological morbidity and mortality. However, data regarding its prevention have not been assorted until date.
OBJECTIVE: The objective of this study was to review the literature and evaluate whether any conclusion can be drawn regarding the effectiveness of any measure on preventing bypass-related CHS.
METHODS: We systematically reviewed PubMed and Cochrane Library from September 2008 to September 2018 to collect data regarding the effectiveness of pharmacologic interventions on the refers to pretreatment (PRE) of bypass-related CHS. We categorized interventions regarding their class of drugs and their combinations and calculated overall pooled estimates of proportions of CHS development through random-effects meta-analysis of proportions.
RESULTS: Our search yielded 649 studies, of which 23 fulfilled inclusion criteria. Meta-analysis included 23 studies/2,041 cases. In Group A (blood pressure [BP] control), 202 out of 1,174 pretreated cases developed CHS (23.3% pooled estimate; 95% confidence interval [CI]: 9.9–39.4), Group B (BP control + free radical scavenger [FRS]) 10/263 (0.3%; 95% CI: 0.0–14.1), Group C (BP control + antiplatelet) 22/204 (10.3%; 95% CI: 5.1–16.7), and Group D (BP control + postoperative sedation) 29/400 (6.8%; 95% CI: 4.4–9.6)].
CONCLUSIONS: BP control alone has not been proven effective in preventing CHS. However, BP control along with either a FRS or an antiplatelet agent or postoperative sedation seems to reduce the incidence of CHS.
Keywords: Bypass surgery, cerebral bypass, cerebral hyperperfusion syndrome, cerebral revascularization, hyperperfusion syndrome, intracranial bypass, reperfusion injury
Cerebral bypass surgery is performed to treat cerebrovascular diseases and tumors involving critical vessels that require either blood flow augmentation (i.e. moyamoya disease [MMD] and steno-occlusive disease [SOD]) or blood flow replacement (e.g. aneurysm and tumor surgery).[1],[2],[3],[4],[5] Advances in neurosurgical technology and microsurgical techniques have enabled a remarkable progression in bypass surgery minimizing patency-related complications under 3%.[6] Nevertheless, cerebral hyperperfusion syndrome (CHS) still remains a major potential complication of cerebral revascularization surgery occurring in rates higher than 40% in single-center series.[7],[8]
CHS can be considered a form of reperfusion injury and is characterized by increased regional cerebral blood flow (rCBF) relative to the homologous area of the contralateral hemisphere and accompanied by nonspecific neurological symptoms.[9] Although bypass-related CHS poses the risk of severe cerebral edema formation followed by irreversible neural damage[10] and life-threatening intracranial hemorrhage,[11],[12] a review of the literature denotes the lack of any assortment of data regarding its prevention (PRE). To that end, we systematically reviewed the literature to assess the effectiveness of PRE interventions and identify interventions that could reduce the risk of bypass-related CHS and ameliorate subsequent neurological deterioration of patients who underwent cerebral revascularization surgery.
MethodsIn this report, we conducted a systematic review on outcome data regarding the effectiveness of pharmacologic interventions used for PRE of bypass-related CHS, according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement.[13] PRE refers to pharmacologic pretreatments, i.e. preventive pharmacologic interventions administered to patients undergoing bypass surgery aiming at bypass-related CHS prevention.
Search methods
We systematically reviewed the PubMed and Cochrane Library databases to identify relevant studies published during the last 10 years (September 2008–September 2018).
Search term conditions
To retrieve relevant articles, we used the following keywords: “hyperperfusion,” “reperfusion,” “bypass,” “revascularization,” “cerebral,” and “intracranial” on the following conditions: ((((hyperperfusion) OR reperfusion)) AND ((bypass) OR revascularization)) AND ((intracranial) OR cerebral).
Search details
((hyperperfusion[All Fields] OR (“reperfusion”[MeSH Terms] OR “reperfusion”[All Fields])) AND (bypass[All Fields] OR revascularization[All Fields])) AND (intracranial[All Fields] OR (“cerebrum”[MeSH Terms] OR “cerebrum”[All Fields] OR “cerebral”[All Fields] OR “brain”[MeSH Terms] OR “brain”[All Fields])) AND ((hyperperfusion[All Fields] OR (“reperfusion”[MeSH Terms] OR “reperfusion”[All Fields])) AND (bypass[All Fields] OR revascularization[All Fields])) AND (intracranial[All Fields] OR (“cerebrum”[MeSH Terms] OR “cerebrum”[All Fields] OR “cerebral”[All Fields] OR “brain”[MeSH Terms] OR “brain”[All Fields])) AND (“2008/09/01”[PDat]: “2018/09/01”[PDat]) [Table 1].
Eligibility assessment
To assess the eligibility of studies (titles/abstracts initially and full texts subsequently), two reviewers (EL and DC) independently reviewed all studies according to the predefined criteria. Results were in agreement, and no additional reviewer was needed to assess disagreements. A detailed schematic representation of the filtering procedure is shown in a flowchart [Supplemental Figure 1].
Data extraction
Initial data extraction was performed according to a predefined data extraction form by three reviewers independently (EL, DC, and DM), and a second data extraction according to PRE was performed by GPS.
Inclusion criteria
We included all clinical studies reporting outcome data regarding the effectiveness of pharmacologic interventions for the PRE of bypass-related CHS [[Table 2] and [Supplemental Table 1] in Supplemental Material, in which inclusion criteria in terms of Participants, Interventions, Comparisons, Outcomes, and Studies are outlined].
Table 2: Inclusion criteria in terms of Participants, Interventions, Comparisons, Outcomes, and Study design
Exclusion criteria
We have excluded studies concerning nonsurgical cerebral revascularization (e.g. endovascular studies) and studies concerning noncerebral revascularization (e.g. coronary and iliac revascularization). Furthermore, we excluded studies written in languages other than English or published outside the last 10-year range. We have also excluded studies that reported pharmacologic interventions for CHS treatment and not for its prevention.
Comparison and outcome definitions
We categorized reported interventions in four PRE groups according to combinations of different classes of drugs used.
PRE groups: Group A – blood pressure (BP) control, Group B – BP control + free radical scavenger (FRS), Group C – BP control + antiplatelet, and Group D – BP control + postoperative sedation.
In our analysis, the outcome was CHS development, namely, to assess the effectiveness of utilized interventions in the context of CHS prevention.
Data analysis
We calculated pooled estimates of the relevant proportions for each PRE group (along with the corresponding 95% confidence intervals [CIs]) through random-effects meta-analyses of proportions and provided exact binomial CIs.[14] Additionally, we performed a test of whether the summary effect measure is equal to zero and a test for heterogeneity, i.e. whether the true effect in all studies is the same. Furthermore, we quantified heterogeneity using the I-squared measure and calculated the pooled estimate of the proportions of interest after a Freeman–Tukey double-arcsine transformation to stabilize the variances.[15],[16] We utilized a random-effects approach in all meta-analyses as the heterogeneity between studies was high and statistically significant in most cases. All analyses were performed using Stata version 14.2 (Stata Corp., TX, USA) and the metaprop command.[17]
Quality assessment of studies
We used the Newcastle–Ottawa Scale for cohorts and case–control studies to assess the quality of included prospective studies as previously described[18] [[Supplemental Table 2] in Supplemental Material]. Moreover, we performed subgroup meta-analyses according to the type of study (retrospective vs. prospective) to compare the quality of data pooled from retrospective studies with quality of data of prospective studies.
Ethics committee approval
It was not applicable as this is a meta-analysis article.
ResultsOur search identified 649 studies. We assessed 135 potentially eligible studies in full text and included 23 studies.[7],[8],[11],[12],[19],[20],[21],[22],[23],[24],[25],[26],[27],[28],[29],[30],[31],[32],[33],[34],[35],[36],[37] A detailed schematic representation of the filtering procedure is shown in a flowchart [Figure 1]. Overall data from 2,041 cerebral bypass surgery cases were included in the meta-analysis. Indications for bypass surgery included MMD in 1,893 cases, SOD in 140, aneurysms in 3, and tumors (e.g. cavernous sinus meningiomas and skull base tumors) in 5 cases. Bypass types used included superficial temporal artery to middle cerebral artery (MCA), internal carotid artery to MCA, and encephalo-duro-arterio-synangiosis [Table 3].
Figure 1: Flowchart schematically demonstrating the filtering procedure of the study selection process according to the PRISMA statement. PRISMA: Preferred Reporting Items for Systematic Reviews and Meta-AnalysesPRE meta-analysis included 23 studies reporting data for 2,041 cases [Figure 2]. In Group A (BP control), 202 out of 1,174 pretreated cases developed CHS (23.3% pooled estimate; 95% CI: 9.9–39.4); in Group B (BP control + FRS), 10 out of 263 pretreated cases developed CHS (0.3%; 95% CI: 0.0–14.1); in Group C (BP control + antiplatelet), 22 out of 204 pretreated cases developed CHS (10.3%; 95% CI: 5.1–16.7); and finally in Group D (BP control + postoperative sedation), 29 out of 400 pretreated cases developed CHS (6.8%; 95% CI: 4.4–9.6). Pooled estimates and 95% CIs from different pretreatment group interventions are demonstrated collectively in a single graph for easier visual comparisons [Figure 3].
Figure 2: Forrest plots for a random-effects meta-analysis of the proportion of pretreated patients who developed CHS by type of study. (a) BP control, (b) BP control + free radical scavenger, (c) BP control + antiplatelet, (d) BP control + postoperative sedation. CHS: Cerebral hyperperfusion syndrome, CI: Confidence interval, ES: Effect size, BP: Blood pressure
Figure 3: Graph showing pooled estimates and 95% CIs of CHS incidence across different pretreatment groups. (A) BP control, (B) BP control + FRS, (C) BP control + antiplatelet, (D) BP control + postoperative sedation. CI: Confidence interval, CHS: Cerebral hyperperfusion syndrome, BP: Blood pressure, FRS: Free radical scavenger Discussion In this systematic review and meta-analysis, we collected and assessed data concerning the effectiveness of drugs utilized for the PRE of bypass-related CHS. Our systematic review confirms the lack of a systematic assortment of relevant data and denotes the extensive variations of utilized pharmacotherapies across studies. Furthermore, our results suggest that BP control alone is not effective for CHS prevention. The addition of FRS or antiplatelet seems to substantially reduce CHS development; likewise, BP control in the setting of postoperative sedation also seems to considerably reduce the incidence of CHS. However, the lack of randomized head-to-head comparison data prohibits the discrimination of treatment effects from study effects; thus, we did not proceed to post hoc tests or meta-regressions among groups. The recently published Japanese Guidelines for the Management of Moyamoya Disease from the Research Committee on Moyamoya Disease and Japan Stroke Society discuss that strict BP control can be effective for the management of symptomatic CHS.[38]
Potential underlying mechanisms of utilized interventions
The neurovascular unit, composed of the endothelium, astrocytes, pericytes, smooth muscle cells, proteins, and enzymes in the extracellular matrix, is considered a major target of reperfusion injury.[39],[40] In patients who underwent bypass surgery, previously long-standing hypoperfused state leads to dysfunction of the neurovascular unit which in turn may disrupt cerebral autoregulation and hence contribute to the appearance of CHS.[30]
More specifically, upsurges of free radicals after reperfusion have been previously shown to impede regional microcirculation by inducing microvessels' pericyte contraction and erythrocyte entrapment despite the recanalization of an occluded artery.[41] Thus, FRS may prevent injury related to free radical byproducts of reperfusion following cerebral ischemia.[23],[29],[31],[34] Edaravone, a widely used FRS, quenches hydroxyl radical (OH●), inhibits OH●-dependent and OH●-independent lipid peroxidation, and enhances NO production without increasing peroxynitrite levels.[42] As a consequence, perioperative administration of edaravone can prevent the occurrence of transient neurological deficits mediated by oxidative radical overproduction.[34] Another agent, minocycline, apart from its antioxidant properties, is suggested to have additional neuroprotective effects, as it inhibits matrix metalloproteinases (MMPs) with concomitant blood–brain barrier maintenance.[23] Considering that MMD patients have increased circulating levels of MMPs and decreased levels of MMP inhibitors,[43] minocycline could be indicated as a pretreatment for MMD undergoing bypass surgery.
The mechanism underlying the neuroprotective effects of antiplatelets in the prevention of bypass-related CHS is not clear. As discussed above, reintroduction of CBF is related to erythrocyte flow obstruction due to pericyte contraction.[41] Hence, it is possible that erythrocyte obstruction enables platelet activation, further compromising regional circulation, and that antiplatelet pretreatment could potentially alleviate this microcirculation obstruction. Furthermore, antiplatelets are reported to be co-administered with antihypertensive agents, for the avoidance of the unfavorable effect of intensive BP lowering – applied for CHS prevention – on the contralateral hemisphere.[22],[23]
Despite not targeting directly the neurovascular unit, postoperative sedation may prevent CHS by means of reduced postoperative rCBF and a more controllable BP.[25],[26],[35] Precisely, continuing propofol or midazolam sedation after surgery permits a better BP regulation when an increase in rCBF is detected and consequently it diminishes the occurrence of postoperative hemorrhagic CHS.[25] The sedation's duration is determined according to postoperative rCBF changes as measured by xenon CT. Therefore, this pretreatment intervention should be reserved only for centers with such diagnostic capabilities or centers which utilize PET or SPECT to measure rCBF.[38]
Cerebral hyperperfusion syndrome prevention in pediatric population: Indications and contraindications
Although previous studies indicate that children are at low risk for CHS development after revascularization surgery,[25],[33],[36],[44],[45] having knowledge of the most suitable PRE interventions in that age group is of great importance, not only to prevent the appearance of CHS but also to avoid hazardous complications of an inappropriate intervention method. More specifically, propofol should not be used as a PRE intervention in children and young adolescents as it has been related to serious adverse effects and the risk of propofol infusion syndrome.[46],[47] Thus, BP control along with a neuroprotective agent or antiplatelet seems to be more prudent choices. However, further studies focused on CHS prevention in pediatric population are needed to support the superiority and elucidate the potential risks of each PRE intervention method.
Suggestions for future studies
Our results indicate the lack of level I evidence for the prevention of cerebral bypass-related CHS and highlight the large variability of the design and utilized pharmacotherapies across studies. Given that CHS development is associated with cognitive deficits,[48] it is important to design prospective randomized controlled trials to investigate differences in effectiveness of Group B, Group C, Group D, or their combinations in head-to-head comparisons. Considering the above, we recommend the following aspects to be considered on the design of future studies assessing cerebral bypass-related CHS.
Future studies should be designed as randomized trials for head-to-head comparisons of potential pretreatments. Furthermore, future studies should report and analyze thoroughly demographics and clinical characteristics of patients undergoing cerebral bypass surgery to facilitate the identification of factors related to CHS development. Specifically, studies should report age, sex, indication for surgery (investigators should clarify duration and severity of ischemia if applicable), date of chronic disease diagnosis (e.g. MMD and SOD), date of acute condition diagnosis (e.g. acute ischemia and hemorrhage), duration between acute condition diagnosis and operation, preoperative prescribed treatment, duration of adherence to preoperative prescribed treatment, pre- and postoperative neuropsychological and cognitive scores, type of bypass, number of anastomoses, clamp time, i.e. intraoperative duration of regional ischemia, pre- and postoperative rCBF measurements, duration between time of operation and onset of symptoms related to CHS, types of symptoms related to CHS, duration of CHS, and duration of pretreatment administration. Finally, investigators should predefine an optimal duration for patient follow-up and report follow-up data accordingly.
Study limitations
This meta-analysis is based on level II–V studies.[49] Systemic biases related with these studies (i.e. selection, detection, performance, attrition, reporting, and publication bias) render their design nonoptimal for assessing the effectiveness of different interventions. Still, these were the only available studies reporting data on pharmacological interventions utilized for the PRE of bypass-related CHS. Although measures used to assess CHS were appropriate and relatively homogeneous across studies, indications for bypass surgery varied from chronic to acute or nonischemic conditions and different types of bypass utilized also varied across studies. CHS development is determined by the condition of the cerebral circulation on the side of the brain that receives blood flow. However, studies included in this meta-analysis do not report any data regarding cerebral circulation before surgery, and this constitutes an additional limitation.
Moreover, we performed meta-analyses according to the type of PRE-utilized interventions which we categorized according to the class of each drug and their combinations. However, as highlighted above, different agents of the same class of drugs may have different biological effect characteristics. Although we compared the effectiveness of CHS prevention of different classes of drugs and their combinations, we were not able to assess the effectiveness of each specific utilized agent due to the very high variability of utilized agents across studies and due to the fact that many studies did not specify the name of utilized agent or utilized more than one agent without specifying outcome results according to agent. Therefore, available data do not permit for agent-specific subgroup analysis.
ConclusionsOur data suggest that BP control alone, as a pretreatment, is not effective for CHS prevention. A FRS or antiplatelet should likely be added as a pretreatment intervention. Alternatively, BP should be controlled in the context of postoperative sedation. Although data are pooled from single-arm studies and therefore pharmacologic effects cannot be distinguished from study-specific effects, we provide data suggesting the added benefit of FRS or antiplatelet or postoperative sedation on BP control for the prevention of bypass-related CHS. Finally, we highlight the need for future randomized trials studying the effectiveness and potential side effects of the above interventions as well as patient characteristics of pretreated patients who developed CHS.
Acknowledgments
We would like to sincerely thank Professors C. Pantos, I. Mourouzis, and A. Lourbopoulos from the Group of Experimental Neurology – Department of Pharmacology, University of Athens School of Medicine, for heartening the group analysis of our data and sharing their pharmacological expertise for defining intervention groups.
Financial support and sponsorship
Nil.
Conflicts of interest
There are no conflicts of interest.
Supplemental MethodsSearch term conditions
To retrieve relevant articles, we used the following keywords to create our search in the PubMed and Cochrane Library: “hyperperfusion,” “reperfusion,” “bypass,” “revascularization,” “cerebral” and “intracranial” on the following conditions: ((((hyperperfusion) OR reperfusion)) AND ((bypass) OR revascularization)) AND ((intracranial) OR cerebral).
Search details
((hyperperfusion[All Fields] OR (“reperfusion”[MeSH Terms] OR “reperfusion”[All Fields])) AND (bypass[All Fields] OR revascularization[All Fields])) AND (intracranial[All Fields] OR (“cerebrum”[MeSH Terms] OR “cerebrum”[All Fields] OR “cerebral”[All Fields] OR “brain”[MeSH Terms] OR “brain”[All Fields])) AND ((hyperperfusion[All Fields] OR (“reperfusion”[MeSH Terms] OR “reperfusion”[All Fields])) AND (bypass[All Fields] OR revascularization[All Fields])) AND (intracranial[All Fields] OR (“cerebrum”[MeSH Terms] OR “cerebrum”[All Fields] OR “cerebral”[All Fields] OR “brain”[MeSH Terms] OR “brain”[All Fields])) AND (“2008/09/01”[PDat]: “2018/09/01”[PDat]).
Eligibility assessment
To assess the eligibility of studies (titles/abstracts initially and full texts subsequently), two reviewers (EL and DC) independently reviewed all studies according to the predefined criteria. Results were in agreement and no additional reviewer was needed to assess disagreements.
Data extraction
Initial data extraction was performed according to a predefined data extraction form by three reviewers independently (EL, DC, DM), and a second data extraction according to PRE was performed by GPS.
Inclusion criteria
We included all clinical studies reporting data regarding the effectiveness of pharmacologic interventions for the PRE of bypass-related CHS. Inclusion criteria in terms of Participants, Interventions, Comparisons, Outcomes, and Studies are detailed in [Supplemental Table 1].
Exclusion criteria
We have excluded studies concerning nonsurgical cerebral revascularization, e.g. endovascular studies, and studies concerning noncerebral revascularization, e.g. coronary and iliac revascularization. Furthermore, we excluded studies written in languages other than English or published outside the last 10 year-range. We have also excluded studies that reported pharmacologic interventions for CHS treatment and not for its prevention.
ReferencesWells G, Shea B, O'Connell D, Peterson J, Welch V, Losos M, et al. Newcastle-Ottawa quality assessment scale cohort studies. Ontario, Canada: University of Ottawa; 2014.Fujimura M, Niizuma K, Endo H, Sato K, Inoue T, Shimizu H, et al. Quantitative analysis of early postoperative cerebral blood flow contributes to the prediction and diagnosis of cerebral hyperperfusion syndrome after revascularization surgery for moyamoya disease. Neurol Res 2015;37:131-8.Fujimura M, Niizuma K, Inoue T, Sato K, Endo H, Shimizu H, et al. Minocycline prevents focal neurological deterioration due to cerebral hyperperfusion after extracranial-intracranial bypass for moyamoya disease. Neurosurgery 2014;74:163-70.Horie N, Fukuda Y, Izumo T, Hayashi K, Suyama K, Nagata I. Indocyanine green videoangiography for assessment of postoperative hyperperfusion in moyamoya disease. Acta Neurochir (Wien) 2014;156:919-26.Uchino H, Kazumata K, Ito M, Nakayama N, Kuroda S, Houkin K. Intraoperative assessment of cortical perfusion by indocyanine green videoangiography in surgical revascularization for moyamoya disease. Acta Neurochir (Wien) 2014;156:1753-60.Teo K, Choy DK, Lwin S, Ning C, Yeo TT, Shen L, et al. Cerebral hyperperfusion syndrome after superficial temporal artery-middle cerebral artery bypass for severe intracranial steno-occlusive disease: A case control study. Neurosurgery 2013;72:936-42.Uchino H, Nakayama N, Kazumata K, Kuroda S, Houkin K. Edaravone reduces hyperperfusion-related neurological deficits in adult moyamoya disease: Historical control study. Stroke 2016;47:1930-2.Fujimura M, Inoue T, Shimizu H, Saito A, Mugikura S, Tominaga T. Efficacy of prophylactic blood pressure lowering according to a standardized postoperative management protocol to prevent symptomatic cerebral hyperperfusion after direct revascularization surgery for moyamoya disease. Cerebrovasc Dis 2012;33:436-45.Kawamata T, Kawashima A, Yamaguchi K, Hori T, Okada Y. Usefulness of intraoperative laser Doppler flowmetry and thermography to predict a risk of postoperative hyperperfusion after superficial temporal artery-middle cerebral artery bypass for moyamoya disease. Neurosurg Rev 2011;34:355-62.Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gøtzsche PC, Ioannidis JP, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: Explanation and elaboration. Ann Intern Med 2009;151:W65-94.References
留言 (0)