Post-stroke infections and preventive antibiotics in stroke: Update of clinical evidence

Abstract

This review provides an update of evidence on post-stroke infections and the use of preventive antibiotics in stroke. Infection is a common complication after stroke, affecting between 15% and 30% of the patients. The predictors for post-stroke infection can be divided into three categories: clinical factors, anatomical (stroke related) factors and immunological factors. The relation between the occurrence of a post-stroke infection and functional outcome remained subject of debate, but it seems likely that the occurrence of these infections has a causal relation with poor functional outcome and mortality. In the first meta-analysis on preventive antibiotic therapy, almost a decade ago, its beneficial effect on post-stroke infection rate was clear; however, the effect on functional outcome remained uncertain because included studies were small and heterogeneous. Afterwards, three large phase-3 RCTs were published and a Cochrane meta-analysis was performed. It has now become clear that, despite the finding that overall infections are reduced, preventive antibiotic therapy in the acute phase of stroke does neither improve functional outcome, nor decrease mortality rates. This does not yet mean that further research on preventive antibiotics in stroke is useless: the pathophysiology and etiology of post-stroke infections are unclear and the use of preventive antibiotics in specific subgroups of stroke patients could still be very effective. This is currently being studied. Besides, preventive antibiotic therapy might be cost-effective by increasing quality-adjusted life years. Thirdly, research for the upcoming years might put more emphasis on the effect of stroke on immunological alterations.

Introduction

Stroke is a main cause of disability and death worldwide, with an estimated 42 million prevalent cases and over 6 million deaths in 2015. ¹ Before this same year, the only proven effective therapy for acute ischemic stroke was intravenously administered recombinant tissue-type plasminogen activator (r-tPA). ² Treatment options were extended in 2015, after multiple large multicenter randomized clinical trials showed beneficial effect of intra-arterial thrombectomy (IAT) when an occlusion occurs in the distal internal carotid or the proximal middle cerebral artery. ³ With the introduction of this second proven treatment option for acute ischemic stroke, it is very likely that in the upcoming years outcome will improve and that mortality rates of stroke will further decrease.

With the aim to further improve outcome, different other therapeutic strategies for acute stroke have been proposed. While the first studies date from 1998, in 2008 the first review suggested that there might be a relation between the occurrence of post-stroke infections and outcome, and that the preventive use of antibiotics could be beneficial. ⁴ In the subsequent years, a possible negative association between post-stroke infections and clinical outcome was reported in different studies.^5–7 The relation between preventive antibiotics and outcome, however, remained unclear. In a Cochrane meta-analysis in 2012, it was found that preventive antibiotics did decrease the post-stroke infection rate, but its effect on functional outcome remained uncertain because the included studies were rather small and heterogeneous. ⁸ Large phase-3 RCTs were needed and since treatment has not yet been proven effective, guidelines on acute stroke management currently state that preventive administration of antibiotics is not indicated. ⁹

Since the publication of this meta-analysis, three large phase-3 RCTs on the effect of preventive antibiotics in acute stroke have been completed.^10–12 As a result, general insights on this subject have markedly changed, as may have recommendations for guidelines and the implications for future research. This review aims to give an update on current knowledge on post-stroke infections and preventive antibiotics.

Post-stroke infections: Prevalence, risk factors, and their impact on clinical outcome

There is debate about the actual prevalence of post-stroke infections. Most literature mainly reports the broad variety in infection rates, ranging between 5 and 65%, because of large heterogeneity of study populations, and definitions for post-stroke infection. Current most complete approximation on the actual post-stroke infection rate is based on a large meta-analysis, including 137,817 patients from 87 different studies with a maximum observation period of 90 days, that showed an overall pooled infection rate of 30% (95% confidence interval [CI] 24–36%). ¹³ Regarding type of infection, the pneumonia rate was 10% (95% CI 9–10%) as was the urinary tract infection rate (95% CI 9–12%).

One could argue, however, that these numbers are overestimated since data are based on selected populations that were prone to detection bias: numbers were derived from randomized controlled trials, cohort studies and stroke registries and detection rate might differ between hospitals with and without stroke research. In a large unselected consecutive cohort study including patients with suspected ischemic stroke, the incidence of a post-stroke infection within seven days was much lower: 15% of the patients had developed ‘any’ overall post-stroke infection. The prevalence for pneumonia was 7.5% and for urinary tract infection 4.4%. ⁷

Microbiologic data on post-stroke infection mainly show an early onset nosocomial pneumonia or a community acquired aspiration syndrome. Staphylococcus aureus and Gram-negative bacteria such as Klebsiella pneumoniae, Pseudomonas aeruginosa, Escherichia coli or Enterobacter spp. are commonly identified, as are Streptococcus species. Pneumonia by aspiration are frequently caused by S. aureus, community acquired pneumonia is often caused by Streptococcus spp. Often, no causative organism is detected in post-stroke pneumonia. ¹³

Predictors for the occurrence of an infection can be divided into three categories: clinical factors, anatomical (stroke related) factors and immunological factors (see also Supplemental Table 1).

Different clinical factors that appear to influence post-stroke infection rate have been described. These can be subdivided into either patient characteristics or factors related to hospital admission. There are general risk factors, like: older age, sex for different types of infection, stroke severity, bedridden state or diabetes mellitus; and factors that directly increase the risk of a certain infection, like: invasive maneuvers such as feeding tube placement or urinary catheterization, mechanical ventilation, presence of chronic obstructive pulmonary disease, dysphagia, or decreased consciousness and reduced bulbar reflexes resulting in aspiration of nasopharyngeal secretions.^13,14

Regarding anatomical (stroke related) factors, to date, only little is known. An association has been suggested between lesion size and the occurrence of infections: patients with larger stroke volume might be at higher risk for pneumonia (odds ratio [OR] 3.5, p < 0.001), compared with patients with smaller stroke volume, and patients with a small lesion have a lower risk for urinary tract infections (OR 0.4, p < 0.01) compared with those with a larger lesion. ¹⁵ Regarding location or even stroke lateralization, no clear association has been found. To our knowledge, only little additional studies have been performed to repeat these findings: two studies also found a small association between stroke volume and the risk of infection,^16,17 whereas another study did not find an independent association between stroke volume and post-stroke infection, but did find a relation with location: an infarction in the anterior middle cerebral artery cortex was significantly associated with the occurrence of infections (OR 5.68; 95% CI 1.00–32.29). ¹⁸

The association between intracerebral hemorrhage volume, localization, and the risk of post-stroke infections remains uncertain as well: one prospective cohort study found an association with a hematoma volume larger than 30cc (OR 2.0; 95% CI 1.1–3.5) as well as a deep location of the hemorrhage (reference lobar: OR 1.90; 95% CI 1.28-2.88), ¹⁹ whereas in two other cohort studies, these associations did not remain significant after logistic regression.^20,21

Thirdly, brain injury itself may influence the risk of a post-stroke infection. It is believed that the central nervous system and the immune system are closely linked and continuously interact via long-distance feedback loops between the central nervous system and peripheral immune organs using the hypothalamo-pituitary-adrenal axis and the sympathetic nervous system. As a result of central nervous system injury, disturbance of these normally well-balanced brain-immune interactions leads to a state called stroke-induced immunodepression.^22–24 The background of this complex interaction is beyond the scope of this review and is yet not fully understood, but in general, brain injury leads to an overactivation of the autonomic nervous system and an increased release of stress hormones resulting in changes in the innate and adaptive immune system. Described features include lymphocyte apoptosis, lymphopenia, altered cytokine production and atrophy of lymphoid organs. ¹⁷

The effect of the occurrence of a post-stroke infection on functional outcome has been discussed in different reviews.^4,14 The majority of studies show an independent association between the occurrence of an infection within the first days after stroke and poor functional outcome and/or mortality.^5,7,25–33 This effect appears to be the particularly present for pneumonia. An overview of the main and most cited studies on this subject has been summarized in Supplemental Table 2.

In general, the occurrence of ‘any’ overall post-stroke infection is associated with an odds ratio for a poor functional outcome that varies between 0.9 and 4.4. The odds ratio for mortality varies between 1.5 and 6.0 (with one outlier with a broad confidence interval ²⁷ ). Despite the fact that different outcome parameters have been used, six of the seven major studies show a significant negative effect of infections on stroke outcome.

Regarding pneumonia, the evidence for functional outcome is limited and odds ratio’s vary with broad confidence intervals of 1.7 to 52. The odds ratio for mortality, however, ranges between 2.1 and 3.0 with small variation between studies. All four major studies showed a significant association.

The effect of post-stroke urinary tract infections on clinical outcome remains subject of debate: of the four major studies looking at this effect, three showed no significant effect.

While the association between infection and a worse functional outcome seems to be proven, the potential mechanisms that could link both parameters remain subject of debate. One could argue that different confounding factors that are more frequent in patients with infections are also predictors of poor outcome, e.g. comorbidity, impaired consciousness and medical interventions. On the other hand, infections could also directly affect outcome via different ways. Firstly, immobility and frailty as a result of a post-stroke infection might lead to a prolonged hospital stay and a delay in rehabilitation. Secondly, infections might induce immunological effects that could worsen outcome. ¹³ Fever as a result of the infection, for example, affects the penumbra, increases the metabolic demands of the cell, exaggerates free oxygen radical production, increases breakdown of the blood–brain barrier and accelerates their transformation into ischemic tissue.^6,34

Preventive antibiotics in stroke: “the early days”

The hypothesis that post-stroke infection might be prevented by antibiotic prophylaxis dates from the 1980s: two studies that were published in 1981 and 1985, respectively, investigated the effect of antibiotic prophylaxis for preventing urinary tract infections in stroke patients in need of an indwelling catheter.^35,36 The first publication on the general prevention of ‘any’ overall post-stroke infection dates from 1982 and investigated the effect of ampicillin and penicillin. ³⁷ The latter study showed no significant effect and after the publication of the aforementioned three manuscripts, it would take years before this subject gained attention again.

The effect of preventive antibiotics on stroke outcome, besides infection rate, gained attention over 20 years later. Starting from 2004, different animal models showed the beneficial effect of, e.g., moxifloxacin, minocycline, and ceftriaxone administered within the first hours after stroke on neurological outcome, mortality, and even infarct size. ⁴ The first large clinical trial, including over 100 patients, was published in 2005 and investigated the effect of intravenous levofloxacin on infection rate, neurological outcome, and mortality. ³⁸ This study also did not show a beneficial effect above optimal care.

Within the next three years, four studies on the effect of preventive antibiotics on infection rate and stroke outcome were published. Although study intervention and duration of therapy varied, these results were compared in a first systematic review and meta-analysis on this subject, published in 2009. ³⁹ There were many major remarks and in general, the study mainly provided a plead for the upcoming large randomized controlled trials on this subject that were to be published within the upcoming years.

A systematic meta-analysis on preventive antibiotic therapy in the acute phase of stroke was published in the Cochrane Library in 2012. ⁸ By that time, five studies involving 506 patients were identified. It was concluded that by the use of prophylactic antibiotic therapy, the incidence of post-stroke infections was significantly reduced (RR 0.58; 95% CI 0.43–0.79), whereas this did not affect stroke outcome, both in mortality (RR 0.85; 95% CI 0.47–1.51) and dependency (RR 0.67; 95% CI 0.32–1.43). The evidence, however, did still not allow a robust and definitive conclusion on the use of preventive antibiotic therapy in acute stroke: the total number of studies and participants was again too limited and there was a broad variety in study design, type of antibiotic therapy (adequately covering all pathogens in post-stroke infections versus mostly chosen for neuroprotective properties), and definitions of infection.

By the end of 2012, after the publication of the first Cochrane meta-analysis, it was concluded that there is not yet evidence that the use of preventive antibiotic therapy should be included in standard care for patients with acute stroke, nor that its potentially beneficial effect should now be rejected. Since included studies were small and heterogeneous, the results warranted evaluation in new stroke trials that needed to enroll large numbers of patients aiming to detect even a small effect, to establish with certainty whether preventive antibiotic therapy has a place in the treatment of patients with acute stroke.

The “new era” of preventive antibiotics in stroke trials

The plead for large clinical trials on the effect of preventive antibiotics on stroke outcome was heard: after the publication of the first Cochrane meta-analysis, three large studies were completed that, in two times, included over the total amount of participants of the former meta-analysis.^10–12

The first large trial was the ‘Preventive Antibiotics in Stroke Study’ (PASS), that was published in The Lancet in 2015. ¹⁰ In this study with a prospective, randomized, open-label, blinded endpoint (PROBE) design, a total of 2538 patients were assigned to receive either intravenous ceftriaxone at a dose of 2 g, given every 24 h intravenously for 4 days, or standard care.

In PASS, the number of infections was significantly reduced (OR 0.55; 95% CI 0.44–0.70), which was mainly driven by a reduction in urinary tract infections (OR 0.34; 95% CI 0.24–0.48), whereas the pneumonia rate did not differ (OR 0.80; 95% CI 0.58–1.10). This study showed no beneficial effect of ceftriaxone on stroke outcome: the distribution of scores on the modified Rankin Scale (mRS) was similar to the control group (adjusted common OR 0.94; 95% CI 0.82–1.09), as was the percentage of patients with an unfavorable outcome (mRS 3-6) when the mRS was dichotomized (OR 0.94; 95% CI 0.80–1.11). Mortality rate at three months follow-up did not differ as well (OR 0.96; 95% CI 0.74–1.24).

The second large trial was the ‘Cluster randomised trial of different strategies of antibiotic use to reduce the incidence and consequences of chest infection in acute stroke patients with swallowing problems’ (STROKE-INF), that was published a few months later in The Lancet as well. ¹¹ In this study with a PROBE design, 1217 stroke patients with swallowing difficulties were enrolled; 48 stroke units in the UK were cluster randomized to preventive antibiotic therapy, in which the choice of the antibiotic was conformed to the local guidelines of participating centers aiming to reduce the incidence of post-stroke pneumonia, or standard care. The incidence of pneumonia, however, was not significantly reduced (OR 1.01; 95% CI 0.61–1.68), nor was the percentage of patients with a good functional outcome increased (mRS 0–2) (OR 0.87; 95% CI 0.6–1.24) or the mortality rate at 90 days reduced (aOR 1.22; 95% CI 0.9–1.64).

The third most recent trial was the ‘Stroke Adverse Outcome is Associated With Nosocomial Infections: PCTus-Guided Antibacterial Therapy in Severe Ischemic Stroke Patients’ (STRAWINSKI) study, that was published in 2017. ¹² In this RCT, 227 patients were randomized to receive either standard stroke care plus Procalcitonin ultrasensitive (PCTus)-guided antibiotic treatment or to standard stroke care alone. Procalcitonin is an early marker of severe bacterial infections and was assessed daily in the intervention group. At a concentration >0.05 ng/ml, a bacterial infection was considered likely and patients were treated with prophylactic antibiotics. Type and duration of antibiotic treatment – aiming to prevent stroke-associated pneumonia – were left to the local guidelines of the participating center. As in the former two studies, the intervention group did not have a better functional outcome on the mRS at 90 days (OR 0.79; 95% CI 0.45–1.39), nor have a lower mortality rate (OR 1.20; 95% CI 0.65–2.24). The overall number of infections, as well as the pneumonia and urinary tract infection rate within the first seven days did not differ as well.

Current insights in the use of preventive antibiotics in the acute phase of stroke

The completion of three recent large clinical trials warranted an update of the Cochrane meta-analysis on this subject. The recent large clinical trials have led to an almost nine-fold increase in study population: from the former five studies involving 506 patients, to eight studies involving 4488 patients. This increase in participants has led to a neutralization of the heterogeneity, making it for the first time possible to draw conclusions on the effect of the ‘overall’ use of preventive antibiotics on functional outcome. The Cochrane meta-analysis has been completely reperformed and was published in January 2018, ⁴⁰ an overview of the characteristics of the included clinical trials has been provided in Supplemental Table 3.

Preventive antibiotics lead to a significant reduction in the general incidence of ‘any’ overall post-stroke infection: from 26% in the control group to 19% (RR 0.71; 95% CI 0.58–0.88). Regarding type of infection, this effect is highly significant for urinary tract infections which were reduced from 10% to 4% in the antibiotic-treated patients (RR 0.40; 95% CI 0.32–0.51), whereas they seem to be unable to prevent pneumonia, which occurs in 11% in the control group versus 10% in the antibiotic group (RR 0.99; 95% CI 0.80–1.13).

In a subanalysis of the studies with the aim of preventing pneumonia by leaving the choice of the antibiotic to local guidelines, there was still no significant difference in the occurrence of pneumonia: 18% in the antibiotics group versus 17% in the control group (RR 1.08; 95% CI 0.87–1.34). As expected, functional outcome did not differ as well.

Regarding functional outcome, prophylactic antibiotics do not have a beneficial effect: the number of patients with a poor functional outcome (dependency or death) do not significantly differ: 55% in the preventive antibiotics group versus 53% in the control group (RR 0.99; 95% CI 0.89–1.10). Mortality is even increased, although not significantly: 17% of the preventive antibiotics group versus 16% in the control group (RR 1.03; 95% CI 0.87–1.21).

No major adverse events are reported. The most reported adverse event was elevated liver and/or renal enzymes, which occurred in 9% of the patients treated with preventive antibiotics. This was also reported in 7% of the patients in the control groups and in none of the patients clinical consequences were reported. Antibiotic resistance, a fearful side effect of the increased use of antibiotics, does currently not seem to be an issue. Infection by a resistant bacterium was reported in 1% of the patients in one study, but happened in the same percentage in the control group. Another study reported the presence of colonization with an antibiotic resistant micro-organism in one patient as well; however, this was already present before the start of the study.

Implications for daily practice

With the completion of the recent large clinical trials, the meta-analysis has gained enough power to state that, despite the heterogeneity, preventive antibiotics do not affect functional outcome after acute stroke and should therefore not be applied in standard stroke care of all patients. The risk of ‘any’ overall post-stroke infection, however, is significantly reduced. This reduction is highly significant for urinary tract infections, whereas no effect on pneumonia is found.

It is considered unlikely that with the completion of new trials in the future this advice will change: in comparison with the former meta-analysis from 2012, point estimates are similar, all confidence-intervals have become smaller and effect sizes on functional outcome and mortality have shifted towards no effect.

Future directions: “Can we conclude that no further research on preventive antibiotics in stroke is necessary?”

Although functional outcome may not be influenced by the use of preventive antibiotics in all stroke patients, it may be false to jump to the conclusion that the research on preventive antibiotics in stroke should now be considered finished. There are different aspects that should be taken into consideration.

First, current data considers the ‘overall’ effect on functional outcome, using ‘any’ type of antibiotic. Although even the large trials and different antibiotic regimens show that functional outcome and mortality are not influenced – even the trials in which the choice of antibiotic was left to local guidelines – data is too limited to draw conclusions on specific antibiotic regimens. It might for instance be possible that the observed beneficial effect of certain antibiotic regimens in animal models is indeed based on neuroprotection instead of the prevention of infections. These results may be neutralized in current meta-analysis.

Second, current data considers the overall effect of antibiotics on ‘any’ patient with stroke. It might be possible that different subgroups of stroke patients do benefit from prophylaxis with antibiotics. In a posthoc analysis, for instance, a beneficial effect was found in thrombolysed patients. ⁴¹ These findings, however, are yet based on a single observation and could still be a finding by chance. Data is currently too limited to make a distinction between ischemic and hemorrhagic stroke, let alone subgroups of stroke patients. Within the next years, data will increase since one international multicenter trial is still ongoing: in the “PREvention of Complications to Improve OUtcome in elderly patients with acute Stroke” (PRECIOUS) study, a targeted 3800 patients will be randomly allocated in a 2 × 2 × 2 factorial design to ceftriaxone, paracetamol, metoclopramide, any combination of these, or to ‘standard’ treatment (ISRCTN82217627).

The next indispensable step within this research field is to identify the patient groups that are most likely to benefit from preventive antibiotic therapy. Based on the hypothesis that the beneficial effect can be explained by the prevention of infections, a prediction rule for post-stroke pneumonia and all post-stroke infections is currently being developed to identify patients at high risk of such infections. ⁴² Data from PASS is used to develop and internally validate these models and, after external validation, these prediction rules could be an important step for the selection of patients for future trials.

Another possibility to further identify subgroups of patients that might benefit from preventive antibiotic therapy is via a pooled analysis of individual patient data. This is currently being performed for the most recent large trials on this subject. Results are expected soon.

The conclusion that there is no role for the use of preventive antibiotics with the aim of improving functional outcome does not exclude that this therapy might still be implied in general practice for its added value in the prevention of post-stroke infections. A recent cost-effectiveness study of PASS showed that three months after stroke, the number of quality-adjusted life years (QALYs) was significantly higher in the ceftriaxone-treated patients, compared with the control group (p = 0.006) and that preventive ceftriaxone in adults with acute stroke is indeed cost-effective. ⁴³ Whether costs justify the increasing use of antibiotics, however, remains subject of debate.

One argument against the more widespread use of antibiotics, especially for a financial reason, might be the occurrence of bacterial resistance. In none of the participants, resistance due to the study intervention has been proven for current studies. The frequent prophylactic administration of antibiotics has already been applied in the intensive care unit by means of selective digestive decontamination (SDD) and selective oropharyngeal decontamination (SOD). One of the aims of this therapy is even to reduce the incidence of ventilator-associated pneumonia. The extensive use of preventive antibiotics in the ICU do not show an increased incidence of colonization or infection with antimicrobial-resistant pathogens. ⁴⁴ Indeed, for all pathogens other than meticillin-resistant Staphylococcus aureus (MRSA), even a lower level of antibiotic-resistance was found. One explanation might be thatthe use of prophylactic selective decontamination antimicrobials could lead to reductions in the need for therapeutic antimicrobials. If the overall net use of antimicrobials is unchanged (or even decreased) with selective decontamination, then there would be no increased antimicrobial selection pressure.

Fourth, current studies have all been performed in Western Europe. It is unknown if current data also apply to other parts of the world, like Asia, where different pathogens are seen and aspects of acute stroke management and strategies for stroke prevention significantly vary. ⁴⁵

Besides arguments that plead for the continuation of the research on preventive antibiotics in stroke, current insights have also led to different new questions that provide implications for future research.

As shown in Supplemental Table 2, pneumonia appears to have the strongest association with poor outcome and mortality of all types of infection. The finding that preventive antibiotics do not influence the post-stroke pneumonia rate, even in the trials in which the choice of antibiotic was left to the local guidelines, might therefore be a plausible explanation for the non-beneficial effect of antibiotics on functional outcome. Indeed, urinary-tract infections can be significantly reduced; however, its effect on stroke outcome is uncertain.

Additional studies to explain this finding are necessary. One suggestion might be that stroke-associated pneumonia is not a bacterial infection, but the result of chemical and immunological alterations that result in a ‘pneumonitis’ that cannot be prevented by antibiotics. ¹⁰ Currently, there is no prove for this finding. However, it has been suggested before that post-stroke infections are only a marker of stroke severity without an independent outcome effect. ²⁹

While research within this field for the last decade mainly drew attention to the effect of preventive antibiotics on infection rate and consequently functional outcome, this might lead to a shift of the main focus for the upcoming years. Instead of infections, research might put more emphasis on the effect of stroke on immunological alterations, e.g. the physiology of stroke-induced immunodepression. If proven that functional outcome may not be influenced by any antibiotic regimen and in any subgroup of stroke patients, the next therapeutic opportunity might become the field of immunological alteration.

Conclusion

Infection is a common complication after stroke. There are different predictors for the occurrence of an infection that can be divided into: clinical factors, anatomical (stroke related) factors, and immunological factors. The majority of reviews show an independent association between the occurrence of a post-stroke infection and poor functional outcome and/or higher mortality. This mainly accounts for pneumonia; the relation between urinary tract infections and stroke outcome is doubtful.

After the completion of three large RCTs that were added to a recent Cochrane meta-analysis, it has now become possible to state that, despite that overall infections are significantly reduced, preventive antibiotic therapy does not improve functional outcome or decrease mortality rates.

This does not yet mean that research on preventive antibiotics in stroke should now be considered finished: a beneficial effect might still be applicable for certain antibiotic regimens and/or subgroups of patients. This is currently being studied. Besides, preventive antibiotic therapy might be cost-effective by increasing the QALYs.

If proven that functional outcome may not be influenced by any antibiotic regimen, in any subgroup of patients, research for the upcoming years might put more emphasis on the effect of stroke on immunological alterations. Consequently, the next therapeutic opportunity to improve stroke outcome might be within immunological alteration.

Supplemental Material

Supplemental Tables - Supplemental material for Post-stroke infections and preventive antibiotics in stroke: Update of clinical evidence

Supplemental material, Supplemental Tables for Post-stroke infections and preventive antibiotics in stroke: Update of clinical evidence by Jan-Dirk Vermeij, Willeke F Westendorp, Diederik van de Beek and Paul J Nederkoorn in International Journal of Stroke