RESEARCH PAPER: Incidence of elevation of cardiac troponin I prior to and following routine general anaesthesia in dogs
Fabio Cilli, Royal Veterinary College, Hawkshead Lane, North Mymms, Herts, AL9 7TA, UK. E-mail: email@example.com
Objective To estimate the incidence of raised cTnI after general anaesthesia in dogs and to explore major risk factors influencing this.
Study design Prospective clinical study.
Animals A total of 107 (ASA physical status 1−2) dogs, 63% male and 37% female, median age 5 years (range 0.3–13.4), median weight 24.4 kg (range 4.2–66.5 kg) undergoing anaesthesia for clinical purposes.
Methods Venous blood samples were taken within 24 hours prior to induction and 24 hours after the termination of anaesthesia. Serum concentrations of cardiac troponin I were measured using a chemiluminescent enzyme immunometric assay with a lower level of detection of 0.20 ng mL−1 (below this level <0.20 ng mL−1). Continuous data were assessed graphically for normality and paired and unpaired data compared with the Wilcoxon signed ranks and Mann–Whitney U-tests respectively. Categorical data were compared with the Chi squared or Fisher’s exact test as appropriate (p < 0.05).
Results Of the 107 dogs recruited, 100 had pre- and post-anaesthetic cTnI measured. The median pre-anaesthesia cTnI was ‘<0.20’ ng mL−1 (range ‘<0.20’–0.43 ng mL−1) and the median increase from pre-anaesthesia level was 0.00 ng mL−1 (range −0.12 to 0.61 ng mL−1). Fourteen dogs had increased cTnI after anaesthesia relative to pre-anaesthesia (14%, 95% CI 7.2–20.8%, range of increase 0.03–0.61 ng mL−1). Six animals had cTnI levels that decreased (range 0.02–0.12 ng mL−1). Older dogs were more likely to have increased cTnI prior to anaesthesia (OR = 5.32, 95% CI 1.35–21.0, p = 0.007) and dogs 8 years and over were 3.6 times as likely to have an increased cTnI after anaesthesia (95% CI 1.1–12.4, p = 0.028).
Conclusion and clinical relevance Increased cTnI after anaesthesia relative to pre-anaesthesia levels was observed in a number of apparently healthy dogs undergoing routine anaesthesia.
General anaesthesia is associated with a relatively high risk of morbidity and mortality in veterinary species, with approximately 1 in 1800 healthy dogs suffering an anaesthetic-related death (Brodbelt et al. 2008). Cardiovascular compromise is an important contributor to mortality and morbidity. Anaesthesia may result in complex changes in cardiovascular function, such as cardiovascular depression with hypotension and reduced tissue perfusion, and in turn to reduced myocardial oxygen delivery and cellular damage. Anaesthesia therefore may represent a potential risk for the myocardium; accurate assessment of myocardial compromise during anaesthesia could help identify those patients that had decreased myocardial oxygen delivery. This may inform strategies to improve oxygen delivery and improve patient management.
If the membranes of myocytes are damaged, the intracellular contents are released into the bloodstream and consequently may be measured by means of specific assay. Cardiac markers were first recommended as part of diagnostic criteria for myocardial infarction by Mahler (1979). Of the available biomarkers, cardiac troponin I (cTnI) is both sensitive and specific for cardiac myocyte ischaemia (Adams et al. 1993; Babuin & Jaffe 2005). Troponin exists as three structural proteins I, C, and T that regulate the calcium-modulated interaction of actin and myosin in striated muscle and there is also a small percentage (<2–8%) unbound in the cytoplasm of the cardiac cells. Troponin I, derived from cardiac and skeletal muscles, is a product of different genes with unique amino acid sequences and hence is readily distinguished (Adams et al. 1993). Due to its cardiac muscle specificity and its very low concentrations in the serum of normal individuals, cTnI has a high sensitivity even for minor levels of myocardial injury. These aspects make cTnI a very powerful biomarker of myocyte injury (Panteghini 2000; Collison et al. 2006). cTnI is highly conserved across species and assays used to detect human cTnI have been validated in the dog (Cummins & Cummins 1987; Schober et al. 1999; Sleeper et al. 2001). Studies in humans have shown that after myocardial cell damage cTnI is released with a biphasic kinetic. Unbound cytoplasmatic cTnI is released within 4–6 hours of myocardial necrosis and reaches a peak concentration at 12–24 hours. Structural cTnI release leads to a second peak 2–4 days after injury (Wolfe Barry et al. 2008; Lee et al. 2009). Recent work comparing changes in cTnI concentrations after anaesthesia with two anaesthetic combinations in dogs did not document significant differences between anaesthetic groups but did report increased cTnI concentrations after anaesthesia in a minority of animals (two of 20 dogs, Saunders et al. 2009). However, further work is merited to estimate with greater precision the frequency of this occurrence. Based on this recent work, we hypothesized that routine general anaesthesia in healthy dogs may result in myocardial hypoxic injury, as indicated by detection of increased serum cTnI. Identification of the frequency of myocardial compromise in healthy dogs undergoing anaesthesia would aid evaluation of routine canine anaesthetic care. Hence, the aims of the study were to evaluate whether general anaesthesia in the dog may result in myocardial injury, as indicated by increased cTnI, and to explore major risk factors associated with myocardial ischaemia.
Materials and methods
Healthy dogs admitted to the Queen Mother Hospital for Animals (QMHA) for clinical procedures (both surgical and non-surgical) necessitating general anaesthesia were recruited after gaining clinician and owner consent between November 2002 and August 2007. The project was approved by the institutional ethics review committee. American Society of Anesthesiologists (ASA physical status) category III–V, patients with suspected or confirmed myocardial disease, and those to remain in the hospital for <24 hours were excluded. A pre-anaesthetic venous blood sample was drawn within the 24 hours of the induction of general anaesthesia. A post-anaesthesia venous blood sample was drawn 24 hours after termination of anaesthesia. Samples were refrigerated (between 2 and 6 °C), centrifuged within 30 minutes, then the serum stored at −20 °C until analyzed. Serum concentrations of cTnI were measured using a chemiluminescent enzyme immunometric assay (IMMULITE Troponin I) designed for use with the IMMULITE Automated Analyzer (DPC, Diagnostic Products Corporation, CA, USA). This assay was validated in humans and has a detection limit of 0.2 ng mL−1 (with values below this level to be referred to as ‘<0.20’ ng mL−1) and a calibration range of up to 180 ng mL−1. The intra-assay variability of cTnI for dogs at this institution was between 3.4% and 4.5% and the inter-assay variability was between 3.1% and 4.2% (English 2009). Details relating to the patient (age, sex, breed, weight, ASA grade), procedure (procedure type and duration) and anaesthetic management (drugs used, monitoring employed and intraoperative cardiopulmonary values) were recorded for all patients. For some dogs, lactate concentrations were measured in venous blood taken at the same time as were the samples for cTnI.
Cardiac troponin I concentrations below the threshold of 0.20 ng mL−1 were recorded as 0.19 ng mL−1 for the purpose of the statistical analysis (though are referred to as ‘<0.20’ throughout the results) and changes in cTnI were calculated as post-anaesthesia cTnI minus pre-anaesthesia cTnI for each patient. The cumulative incidence and 95% confidence interval (95% CI) for increasing cTnI after anaesthesia (relative to pre-anaesthesia concentrations) were calculated by standard methods (Kirkwood 1988). Continuous data were assessed graphically for normality. Pre-anaesthesia cTnI concentrations were compared to post-anaesthesia cTnI with the Wilcoxon signed ranks test. Patients with an increase in cTnI were compared to those without an increase for the patient, procedure and anaesthetic management factors recorded above, using the Mann–Whitney U-test for continuous data and the Chi squared test or Fisher’s exact test for categorical variables as appropriate. All analyses were undertaken with standard statistical software (spss version 17.0 and Stata 9.0; Statacorp, IL, USA). Significance was set at the 5% level.
One hundred and seven healthy (ASA physical status 1–2) dogs were included in the study. The median age of dogs was 5 years (range 0.3–13.4), median weight was 24.4 kg (range 4.2–66.5 kg) and 63% were male and 37% female. German Shepherd Dogs were the most commonly represented breed (12.2%), followed by cross breeds (8.4%), Springer Spaniels, Staffordshire Bull Terriers and Labradors (6.5% each). Procedures undertaken included diagnostic ‘workups’ (25.8%), soft tissue (20.3%), orthopaedic (35.2%) and neurosurgery (2.3%). The median duration of anaesthesia was 140 minutes (range 15–450 minutes). Dogs were premedicated most frequently with acepromazine (ACP injection; Novartis, UK) and an opioid (75/107; 71% of dogs) or with an opioid only (20/107; 19% of dogs). Occasionally dogs were premedicated with medetomidine (Domitor; Pfizer Animal Health Ltd, UK) combinations (9/107;8.6% of dogs) and one dog received acepromazine only (1/107;1%). Two dogs had missing information and therefore their premedication was unknown. Anaesthesia was induced with thiopental (Thiopental injection BP; Link Pharmaceutics Ltd, UK, 64/107; 60% of dogs), or with propofol (Vetofol injection; Norbrook Laboratories Ltd, UK, 42/107; 40% of dogs). For one dog the induction agent was unknown. The inhalation agents used for anaesthetic maintenance were isoflurane (Isoflo; Abbot Animal Health Laboratories Ltd, UK) (71/107; 64% of dogs), halothane (Halothane-Vet; Merial Animal Health Ltd, UK, 33/107; 31%), sevoflurane (SevoFlo; Abbot Laboratories Ltd, UK 4/107; 3.8%) and desflurane (Suprane); Baxter Health Care Ltd, UK, 1/107; 1%). Two dogs had two inhalation agents used during the same anaesthetic. All but three dogs received fluid therapy during anaesthesia.
The median preanaesthesia cTnI concentration was ‘<0.20’ ng mL−1 (range ‘<0.20’ to 0.43 ng mL−1). Of 105 dogs with pre-cTnI concentrations available, where a record of health status was available (n = 64), healthy dogs (ASA grade 1) had a median pre-anaesthesia cTnI concentration of ‘<0.20’ ng mL−1 (range ‘<0.20’–0.43 ng mL−1, n = 40) and dogs with mild to moderate systemic disease (ASA 2) had a median pre-anaesthesia cTnI concentration of ‘<0.20’ ng mL−1 (range ‘<0.20’–0.36 ng mL−1, n = 24). The median post-anaesthesia cTnI concentration was ‘<0.20’ ng mL−1 (range ‘<0.20’–0.80 ng mL−1), and the median increase was 0.00 ng mL−1 (range −0.12 to 0.61 ng mL−1) (Table 1). There were 12/105 dogs with cTnI above the level of detection (≥0.20 ng mL−1) before anaesthesia (six of which increased and six decreased after anaesthesia) and 21/102 had post-anaesthesia cTnI concentrations above the level of detection (14 increased, five decreased and two were unknown (no pre-sample) relative to their post-anaesthesia samples). Fourteen dogs, out of the 100 that had both pre- and post-cTnI samples, had increasing cTnI concentrations after anaesthesia relative to their pre-anaesthesia cTnI concentrations (14%, 95% CI 7.2–20.8%) with increases ranging from 0.03 to 0.61 ng mL−1. Six animals had cTnI levels that decreased after anaesthesia (relative to their pre-anaesthesia cTnI concentrations, range 0.02–0.12 ng mL−1) (Table 2). Their pre-anaesthesia cTnI concentrations ranged from 0.22 to 0.43 ng mL−1. None of these dogs had an obvious history of a recent cardiac insult prior to anaesthesia.
Table 1. Cardiac troponin I (cTnI) serum values prior to and after anaesthesia in dogs. ‘<0.20’ ng mL−1 indicates values below the level of assay detection
|Pre- anaesthesia||‘<0.20’||‘<0.20’ to ‘<0.20’||‘<0.20’ to 0.43||105|
|Post-anaesthesia (24-hour sample)||‘<0.20’||‘<0.20’ to ‘<0.20’||‘<0.20’ to 0.80||102|
|Difference||0.00||0.00–0.00||−0.12 to 0.61||100|
Table 2. Details of dogs for which cardiac troponin I (cTnI) increased after anaesthesia. ‘<0.20’ ng mL−1 indicates values below the level of assay detection
|German Shepherd||1.92||M||Total hip replacement||350||‘<0.20’||0.43||0.24|
|Flat Coat Retriever||8.25||M||Limb osteosarcoma amputation||130||‘<0.20’||0.80||0.61|
|Shetland Sheep Dog||2||M||TPLO stifle surgery||300||‘<0.20’||0.36||0.17|
|Staffordshire Bull Terrier||10||F||Patella repair||205||‘<0.20’||0.43||0.24|
|Staffordshire Bull Terrier||5||M||Cruciate ligament repair||165||0.34||0.37||0.03|
|Dogue de Bordeaux||1||F||TPLO stifle surgery||380||‘<0.20’||0.36||0.17|
|Weimaraner||9||M||MRI and CSF tap||120||0.27||0.37||0.10|
|Bichon Frise||11||M||Perineal hernia repair||140||‘<0.20’||0.32||0.13|
|Bulldog||8||M||CT, X-rays and ultrasound||110||0.34||0.71||0.37|
|Bernese Mountain Dog||10||M||MRI scan||55||0.35||0.49||0.14|
|Doberman||6||M||TPLO stifle surgery||205||0.36||0.44||0.08|
|English Springer Spaniel||8||M||Perineal hernia repair||150||‘<0.20’||0.30||0.11|
There were no significant differences in the number of animals that had recorded reduced or elevated heart rates (<60 beats minute−1, p = 0.90, >150 beats minute−1, p = 0.67 respectively), presence of ventricular premature complexes during anaesthesia (p = 0.18), elevated mean arterial blood pressure (>120 mmHg, p = 0.21), low arterial oxygen saturation (<92% arterial saturation, p = 1.00) between those dogs that had an increase in cTnI concentration (postoperative compared to preoperative) and those that did not.
Of the risk factors evaluated, only animal age was associated with an increase in cTnI after anaesthesia. Dogs of 8 years and over were 3.6 times as likely to have an increased cTnI after anaesthesia (post-anaesthesia compared to pre-anaesthesia cTnI concentrations, 95% CI 1.1–12.4, p = 0.028). There was no significant difference between the median age in dogs demonstrating an elevated cTnI after anaesthesia compared to dogs that did not (cTnI increased post-anaesthesia compared to pre-anaesthesia). However the median age tended to be higher in dogs which showed an increase in cTnI post-anaesthesia (median age 8 years IQR 5–9) compared to the dogs that did not (median age 4 years IQR 1.8–7.6, p = 0.061). However, old dogs were more likely to have an elevated cTnI prior to anaesthesia than young dogs (OR = 5.32, 95% CI 1.35–21.0, p = 0.007). There were no differences in sex between those that had an increase in cTnI and those that did not (8% of females, and 17% of males, p = 0.243) and weights of dogs were similar for those that had a raised cTnI (median 31.1 kg (IQR 17.8–44.5) and those without an increase (median 24.0 kg (IQR 14.8–33.5) p = 0.307). There was no significant difference in the frequency of long procedures (120 minutes or greater, p = 0.189) for dogs with increased and non-increased cTnI after anaesthesia. There were no associations between premedication, induction or maintenance agents used and an increase in cTnI. Intraoperative extremes of heart rate, mean arterial blood pressure (MAP) and oxygen saturation levels were not different between dogs with raised cTnI and those with no increase (see Table 3). There was no major change in serum lactate after anaesthesia compared to pre-anaesthesia serum lactate. All dogs recovered uneventfully from anaesthesia.
Table 3. Intraoperative cardiovascular and other parameters for dogs in which cardiac Troponin I (cTnI) increased following anaesthesia, and for dogs in which cTnI did not increase
|Mean heart rate (beats minute−1)|| 99 (72 to 155)||8|| 88 (47 to 155)||54||0.796|
|Lowest heart rate (beats minute−1)|| 76 (44 to 120)||8|| 66 (33 to 120)||54||0.619|
|Highest heart rate (beats minute−1)||138 (84 to 181)||8||117 (54 to 192)||54||0.207|
|Mean MAP (mmHg)|| 86 (75 to 100)||5|| 85 (48 to 118)||25||0.125|
|Lowest MAP (mmHg)|| 72 (62 to 88)||5|| 73 (32 to 103)||25||0.476|
|Highest MAP (mmHg)||101 (80 to 131)||5||100 (65 to 160)||25||0.721|
|Lowest SpO2 (%)|| 88 (69 to 97)||7|| 93 (60 to 98)||47||0.672|
|Lowest body temperature (°C)|| 35.5 (34.1 to 36.3)||7|| 36.0 (31.7 to 37.4)||36||0.420|
|Lactate change after anaesthesia (mmol L−1)|| −0.45 (−1.4 to 0.4)||8|| 0.2 (−2.3 to 1.3)||19||0.067|
This study documents a frequency of increasing cardiac troponin I after anaesthesia of 14% for healthy dogs presenting for routine procedures, suggesting minor myocardial cell damage occurred in these dogs during or soon after anaesthesia. The frequency of elevated cTnI documented here was consistent with previous work in a smaller group of animals in which 10% (two out of 20) of dogs experienced elevated cTnI after routine surgery (Saunders et al. 2009). In most dogs intraoperative events were documented including haemorrhage, periods of hypoxia (SpO2 < 92%), extremes of heart rate and arterial blood pressure and these disturbances may have contributed to the cTnI increases detected. However a number of these events also were reported in those dogs that did not show an elevation in cTnI after anaesthesia and hence it is not possible to confirm that these intraoperative events were associated with the increases in cTnI detected. Nonetheless, the changes in cTnI detected were consistent with minor cardiac insult and suggest that even after apparently minor physiological disturbance during anaesthesia in healthy dogs, some cardiac ischaemia may occur.
Cardiac Troponin I (cTnI) has been documented previously as a sensitive and specific marker of cardiac ischaemia in humans and animals (Adams et al. 1993; Sleeper et al. 2001; Babuin & Jaffe 2005; Hagman et al. 2007). Previous work at this institution using this assay has reported coefficients of variability in dogs of between 3.4% and 4.5% for intra-assay and between 3.1% and 4.2% for inter-assay variability (English 2009), suggesting all increases detected in the current study were unlikely to represent variation in assay performance. Further, other work using the same assay has also documented low cTnI concentrations in normal dogs with ranges reported of between <0.2 and 0.3 ng mL−1 and <0.05–0.24 ng mL−1, suggesting the majority of elevated and increasing cTnI concentrations detected were associated with myocardial cell damage (Spratt et al. 2005; Hagman et al. 2007). Using other cTnI assays, normal canine ranges have been reported to be even lower (0.03–0.07 ng mL−1, Sleeper et al. 2001; <0.1–1.1 ng mL−1, Burgener et al. 2006). It is important to caution comparisons between values of cTnI measured with different assays as large variations in values between analysers have been reported (Adin et al., 2005). Hence, it would appear that the increases detected were greater than the limits of measurement error and were likely to reflect genuine myocardial cell damage.
For the six dogs with a decrease of post-operative cTnI, pre-anaesthetic myocardial damage was hypothesized such that the measured cTnI concentrations after anaesthesia may have decreased if the myocardium was recovering. The files of these patients were evaluated retrospectively but no signs of myocardial injury before the anaesthetic were confirmed by the patient’s history. Based on the IMMULITE test performance and careful handling undertaken of all samples, the authors consider these results unlikely to be the result of variation in assay measurement or sample handling, though a clear explanation is not apparent.
The increased concentrations of cTnIs detected were consistent with compromise to the myocardium but they were unlikely to reflect major myocyte death. In humans, reduced oxygen supply, increased myocardial oxygen consumption, increased wall stress, tachycardia, hypertension with left ventricular hypertrophy and cardiac trauma have all been suggested to induce myocardial injury (Gunnewiek & Van Der Hoeven 2004) and though not documented in dogs, these changes are likely to also underlie canine myocardial damage. A clear threshold above which major damage is likely has not been established in dogs. Previous work has documented large increases in cTnI when major myocardial damage was likely to have occurred. Dogs with gastric dilatation volvulus had cTnI concentrations of 2.05–24.9 ng mL−1 (Schober et al. 2002) and dogs with pyometra had cTnI values of 0.3–0.9 ng mL−1 (Hagman et al. 2007). Previous work evaluating cTnI concentrations in dogs with cardiac disease reported a wide spectrum of values: ranging from 0.03 to 1.88 ng mL−1 for dogs with cardiomyopathy, 0.01–9.53 ng mL−1 for degenerative mitral valve disease, and 0.01–0.94 ng mL−1 for subvalvular aortic stenosis (Oyama & Sisson 2004). In the current study, no increase >0.61 ng mL−1 was detected suggesting only minor myocardial cell damage occurred. In humans a cutpoint value of cTnI has not been defined yet, but recent work has demonstrated that even small increases in cTnI, were associated with poorer prognosis (Wolfe Barry et al. 2008). Hence, the current results are consistent with compromise to cardiac oxygen supply during anaesthesia, and may reflect minor adverse affects on patient myocardial function.
That none of the increases detected was severe may reflect that all these patients were closely managed and monitored during anaesthesia. All patients were routinely monitored with the measurement of heart rate, arterial blood pressure, capnography and pulse oximetry and the presence of this monitoring may have allowed the attending anaesthetists to treat any developing physiological imbalances before more serious sequelae ensued and hence this may reflect the only minor increases in cTnI detected. In a recent epidemiological study of small animal anaesthetic deaths, the use of pulse oximetry in cats was associated with reduced odds of anaesthetic death (Brodbelt et al. 2007). It is likely that early detection of compromise to patient tissue oxygen delivery can reduce complications and close monitoring of patients should continue to be recommended.
When evaluation of possible risk factors for raised cTnI was undertaken, only increasing age was detected as a risk factor. However, old dogs were also more likely to have elevated pre-anaesthetic cTnI concentrations and it may be that though changes in cTnI were similar for old and young dogs, increases after anaesthesia in young dogs may have been more likely to remain below the level of assay detection. The use of a more sensitive assay would be advocated to allow detection of even more minor changes below this level of detection. Previous work has also suggested older dogs may be at increased risk of occult myocardial disease and the current results may reflect this also (Oyama et al. 2007). That no drug associations, intraoperative parameters or other risk factors were associated with an increase in cTnI may reflect the limited power of the current study to detect risk factors rather than a genuine lack of association. Additional work is required to evaluate risk factors further.
This study was subject to a number of limitations. There were a relatively small number of animals with increased cTnI after anaesthesia and as such it would have had limited power to detect risk factors for increased cTnI. Nonetheless, 14% of patients had an increase of cTnI after anaesthesia suggesting that minor myocardial cell damage may have occurred during routine anaesthesia. Some data were not available, due to the inability to locate the written anaesthetic record for all patients and this reduced the study power and the ability to evaluate risk factors. Though the peak increase in cTnI has previously been reported to occur 12–24 hours after myocardial damage, some elevation could have occurred before this time and by sampling at 24 hours after termination of anaesthesia in the current study, the maximum rise may not always have been detected (Burgener et al. 2006; Wolfe Barry et al. 2008; Lee et al. 2009). However, the decline in cTnI after an insult has been reported to take 2–5 days and hence in the current study it is likely that elevations per se after anaesthesia would have been detected (Colantonio et al. 2002; Burgener et al. 2006; Lee et al. 2009). Further, the population studied was derived from a referral hospital and though they were all classified as healthy (ASA 1–2), the nature and duration of procedures undertaken were likely to be more complex and longer than those undertaken in primary practice. As such the risk of raised cTnI after anaesthesia reported of 14% may be greater than that experienced in first opinion practice. Further work, evaluating changes in cTnI in primary practice would be merited. In addition, though no heart disease was detected on clinical examination, the possibility of occult heart disease being present in some patients cannot be excluded, however the effect of this might be primarily to exacerbate any increase in cTnI detected as opposed to increase the frequency of increased cTnI per se (Oyama et al. 2007). Finally, the sensitivity of the assay was only moderate, suggesting minor elevations after anaesthesia in some dogs may not have been detected due to the pre- and post-anaesthetic samples remaining below 0.20 ng mL−1, and though the current work identified elevations that are likely to be clinically relevant, further work using a test that is able to detect lower levels of cTnI would be merited.
In conclusion, this study documents a frequency of raised cTnI after anaesthesia of 14% in healthy dogs. Older patients were at increased risk of raised cTnI and greatest care could be exercised when anaesthetising old animals to ensure optimal myocardial oxygen delivery. The measurement of cTnI would appear useful in assessing intraoperative myocardial cell damage and additional work is merited to identify the cause of cardiac cell damage during anaesthesia.
The authors would like to acknowledge the generosity of Petplan in supporting this project. From the Royal Veterinary College we would like to thank the Clinical Investigation Centre and clinicians from the Queen Mother Hospital for Animals for their help and the clinical pathology team for the measurement of cTnI concentrations. We would also like to thank the owners of participating dogs.