Retrospective evaluation of 22 dogs with leptospirosis treated with extracorporeal renal replacement therapies (2018‐2021)

Abstract Background Outcomes of dogs with acute kidney injury secondary to leptospirosis (AKI‐L) treated using renal replacement therapies (RRT) are poorly characterized. Hypothesis/Objectives Describe survival to discharge, short (≤30 days) and long‐term (≥6 months) outcomes of AKI‐L dogs receiving RRT and determine if there is a significant difference in maximum blood urea nitrogen (maxBUN), maximum creatinine (maxCr), maximum bilirubin (maxBili) and the number of body systems affected between survivors and non‐survivors. Animals Twenty‐two client‐owned dogs with AKI‐L receiving RRT. Methods Retrospective medical record review of dogs with AKI‐L that received RRT between 2018 and 2021. Results Sixteen of 22 (73%) dogs survived to discharge. Of the survivors, 13 (81%) were alive >30 days from discharge and 12 (75%) were alive at 6 months from discharge. Factors significantly higher in non‐survivors included number of body systems affected (survivors: 1 (19%), 2 (50%), 3 (25%) and 4 (6%) vs non‐survivors: 3 (33.3%), and 4 (66.7%); P = .01) and median maxBili (survivors: 1.9 mg/dL; range, 0.1‐41.6 vs non‐survivors: 21.0 mg/dL; range, 12.3‐38.9; P = .02). There was no significant difference in median maxBUN (survivors: 153.0 mg/dL; range, 67‐257 vs non‐survivors: 185.5 mg/dL; range, 102‐218; P = .44) and median maxCr (survivors: 9.8 mg/dL; range, 6.2‐15.9 vs non‐survivors: 9.8 mg/dL; range, 8.4‐13.5; P = .69) between survivors and non‐survivors. Conclusions and Clinical Importance Regardless of azotemia severity, dogs with AKI‐L receiving RRT have a good survival rate to discharge. The number of body systems affected and hyperbilirubinemia might be associated with worse outcomes.


| INTRODUCTION
Leptospirosis is a widespread, worldwide zoonotic disease caused by spirochetes of the genus Leptospira. 1 In dogs, leptospirosis results in inflammatory, multisystemic disease, predominantly affecting the kidneys and liver. 2,33][4] Survival depends on clinical sign severity and secondary complications. 3,5[7] During the disease's acute phase, dogs typically are presented with an AKI secondary to acute interstitial nephritis. 8Thirty percent of dogs with acute leptospirosis develop oliguria or anuria. 4These dogs might die before they reach the recovery stage of kidney function and therefore extracorporeal renal replacement therapy (RRT) is recommended. 9This therapy can remove uremic toxins, correct fluid and electrolyte disorders and restore acid-base balance until the recovery phase begins. 9A survival rate of 86% is reported in dogs with an AKI-L that received treatment with intermittent hemodialysis (IHD). 5[12] There is a need for updated information on the survival of dogs with AKI-L when RRT is available, and on factors that might affect their prognosis.
The aim of this retrospective study was to describe the survival to discharge, short-term (≤30 days from discharge) and long-term (≥6 months from discharge) outcome in dogs with AKI-L requiring RRT.A secondary aim was to determine if maximum blood urea nitrogen (maxBUN), maximum serum creatinine (maxCr), maximum bilirubin (maxBili) and the number of body systems affected might be associated with survival to discharge.

| MATERIALS AND METHODS
The medical records of dogs diagnosed with leptospirosis and treated with RRT between January 2018 and December 2021 at Tufts Cummings School of Veterinary Medicine (TCSVM) were retrospectively reviewed.

| Inclusion and exclusion criteria
Inclusion criteria included a complete medical record, a confirmed diagnosis of leptospirosis and treatment with at least 1 session of RRT.
Criteria for leptospirosis diagnosis included a single micro agglutination test (MAT) titer >1:800 for nonvaccine serovars or >1:1600 for vaccine serovars, a greater or equal than 4-fold increase in convalescent titers performed 14 days after original titer, or a positive urine or blood polymerase chain reaction (PCR) test.Exclusion criteria included incomplete medical records, dogs not diagnosed with leptospirosis using the criteria above, and dogs with AKI-L not requiring RRT.

| Diagnostics and hospitalization
The extracorporeal therapy database at TCSVM was screened for dogs who received RRT and were tested for leptospirosis between 2018 and 2021.Dogs with a confirmed leptospirosis diagnosis had medical records at TCSVM and referring veterinarian records reviewed for signalment, presenting clinical signs, clinical parameters (heart rate, respiratory rate, temperature, weight), hydration status and urine production.
Fluid overload was clinically assessed based on >10% weight gain from euhydrated baseline, chemosis, peripheral edema, ascites, pleural effusion, pulmonary edema or serous nasal discharge. 13Urine output (UOP) was classified as polyuria (>2 mL/kg/h), oliguria (<1 mL/kg/h) and anuria (no urine production >6 hours) 14 and was measured before RRT by either referring veterinarians or via an indwelling urinary catheter at our institution.Vital parameters were evaluated to assess for persistent tachypnea (>40 breaths per minute) or dyspnea.Dogs that required transfusions were first blood typed then transfused with type-specific blood.Dogs that had received a transfusion >72 hours prior had a major cross match performed and were transfused with a compatible blood product.
Complete blood count (CBC), serum biochemistry and urinalysis within 24 hours of RRT initiation were reviewed.Increased serum creatinine (sCr) was defined as sCr ≥1.6 mg/dL, increased serum BUN as BUN ≥30 mg/dL and AKI was defined based on IRIS AKI grading system. 14 Leptospirosis Society standards 15 and measured serum antibodies against Leptospira interrogans serovar Autumnalis, Bratislava, Canicola, Icterhaemorrhagiae, Pomona and Leptospira kirschneri serovar Grippotyphosa.The urine or ethylenediamine tetra-acetate anticoagulated blood PCR was performed according to World Organization for Animal Health (OIE) standards. 16A confirmatory diagnosis was made based on MAT and/or PCR testing.

| Extracorporeal renal replacement therapy
Dogs were considered RRT candidates if they had a documented AKI with 1 or several of: inadequate urine production, hyperkalemia, fluid overload or progressive azotemia that was unresponsive to medical management.
All dogs had a size-appropriate (range, 8 Fr Â 12 cm to 14 Fr Â 30 cm) double-lumen temporary hemodialysis catheter placed in an external jugular vein using the modified Seldinger technique while under deep sedation or intubated under anesthesia.Intermittent hemodialysis sessions (IHD) were performed on a Fresenius 2008T platform (Fresenius Medical Care, Bad Homburg, Germany).A hybrid, intermittent version of diffusive and convective treatment using a continuous renal replacement therapy (CRRT) platform was performed on the Prismaflex platform (Gambro Lundia AB, Lund, Sweden) if the Fresenius 2008T platform was unavailable.The dialyzer was chosen based on patient size, degree of azotemia and priming volume.Priming was performed with 0.9% NaCl as the priming solution.For our most recent patient, a cytokine reduction filter, VetResQ (CytoSorbents, Inc, Monmouth Junction, New Jersey) was added pre-dialyzer in series to the IHD circuit.All dogs received systemic anticoagulation with unfractionated heparin (UFH) or regional citrate anticoagulation (RCA).Dogs anticoagulated with UFH were monitored by activated clotting time (ACT) (ACT Plus Automated Coagulation Timer System, Medtronic, Minneapolis, Minnesota) to maintain the ACT within an acceptable range of 180 to 200 seconds.Dogs with a clinically relevant thrombocytopenia (<60 k/μL), were anticoagulated with RCA due to the bleeding risk associated with systemic anticoagulation.In accordance with a previously published protocol, 17 4% trisodium citrate was added to the circuit as the blood was withdrawn from the dialysis catheter's arterial port to lower the iCa in the extracorporeal circuit to a target range of 0.2 to 0.4 mmol/L.Just before the blood was returned to the patient, 10% calcium gluconate was added via the dialysis catheter's venous port to target an iCa of 0.8 to 1.2 mmol/L in the patient.
The dialysis prescription was adjusted to the needs of individual patients, according to published protocols. 18The target urea reduction ratio (URR) was determined based on the serum BUN before RRT and on the number of treatments in accordance with previously established safe URR targets. 18Ultrafiltration was performed in fluidoverloaded dogs and ultrafiltration rate was based on the dog's clinical status.Monitoring and adjustments of ultrafiltration were based on the percent change in the dog's intravascular blood volume displayed on the Crit-Line III monitor (Fresenius Medical Care, Waltham, Massachusetts).All dogs were monitored with a 6-lead EKG, temperature probe and blood pressure monitoring.
Session records were reviewed for complications during individual RRT sessions.Complications were characterized as not requiring intervention, requiring emergent intervention and requiring urgent session termination.Dialysis disequilibrium was defined as a central nervous system disturbance caused by a rapid decrease in serum osmolality due to a highly effective RRT session. 19

| Body systems affected
Dogs were grouped by the number of body systems affected based on the target organs of leptospirosis. 2,3Body systems affected were determined based on CBC, serum biochemical profiles, urinalysis, coagulation testing, imaging and concurrent clinical signs.Kidney involvement was defined according to the IRIS AKI grading scheme. 14Hepatic involvement was defined as elevations in 2 or more serum liver enzyme activity, with or without hyperbilirubinemia.Dogs with ultrasonographic pancreatic or gallbladder abnormalities without increased liver enzyme activity were not considered to have hepatic involvement.Pulmonary involvement was defined as persistent tachypnea and dyspnea with either evidence of an alveolar or interstitial pattern on thoracic radiographs or evidence of lung disease on necropsy.Coagulation system involvement was defined as abnormal results of coagulation testing before RRT, clinically relevant thrombocytopenia (minPLT <60 k/μL) or clinical evidence of thrombi or bleeding tendencies (epistaxis, petechial hemorrhages).

| Outcomes
Survival to discharge was defined as dogs alive at hospital discharge.
Serum creatinine at discharge was recorded for surviving dogs and before death for non-surviving dogs.Assessment of short-term survival (≤30 days) and long-term (≥6 months) survival were performed by reviewing records at TCSVM and from referring veterinarians to determine whether a dog that survived to discharge was alive or dead at 30 days and at 6 months from discharge.Dogs alive after 6 months from discharge, were staged according to IRIS CKD Staging guidelines 20 based on their most recent sCr.

| Statistical analysis
Data were tested for normality using histograms.For continuous variables, median and ranges were computed.For noncontinuous variables, frequencies and percentages were reported.To determine if there were differences based on survival status, Wilcoxon rank sum tests and Fisher's exact test (for noncontinuous variables) were employed due to low numbers of data points.Spearman correlation analysis was used to assess the association between maxBili and maxBUN and maxCr.All analyses were conducted using RStudio v.4.2.2 21 and 0.05 was used as a level of significance.No dogs had positive results for Canicola.

| Clinical presentation
Twenty-one dogs (95%) were referred for RRT.Before referral, 20 dogs (95%) were hospitalized at another hospital for a median of 2 days (range, 1-4).Eighteen dogs (82%) had received antibiotic apy before presentation.One dog had received treatment for hypertension with amlodipine.
Eighteen dogs (82%) had elevations in all or 1 of the following: alkaline phosphatase (ALP), alanine transaminase (ALT), aspartate transaminase (AST), gamma-glutamyl transferase (GGT) and bilirubin (see Table 1).On intake, 14 dogs (64%) were hyperbilirubinemic with a median serum bilirubin of 2.5 mg/dL (range, 0.1-23.9mg/dL).In most dogs (77%), bilirubin continued to rise with a median maxBili concentration of 8.0 mg/dL (range, 0.  Two dogs had echocardiograms.One dog was diagnosed with a significant, malignant ventricular arrhythmia secondary to hypokalemia (2.3 mEq/L; RR: 3.4-4.9)secondary to severe diuresis during the recovery stage.One dog was diagnosed with left ventricular hypertrophy and a suspected thrombus in the right atrium.
All dogs were assessed for liver involvement based on elevations in 2 or more serum liver enzyme activities and bilirubin.Sixteen dogs (73%) had hepatic involvement.Of those, 9 dogs had abnormalities on hepatic imaging which included a hypoechoic liver (9) and hepatomegaly (5).Three dogs also had a hypoechoic pancreas and 7 had gallbladder sludge on abdominal ultrasound.
All dogs were assessed for pulmonary involvement based on persistent dyspnea and tachypnea.Twenty dogs had radiographs performed.Nine dogs had clinical signs of respiratory involvement and radiographic changes.One dog had severe dyspnea and tachypnea without radiographic changes.On necropsy, the dog had lung changes with diffuse, neutrophilic interstitial pneumonia with perivascular edema.These 10 dogs (45%) were classified as having pulmonary involvement.Of the 20 radiographed dogs, 10 (50%) had radiographic abnormalities including an alveolar pattern ( 6) with or without an interstitial pattern (10).No other dogs had clinical evidence of dyspnea and tachypnea without radiographic changes.
All dogs were assessed for coagulation involvement by documenting clinical evidence of an hemostatic disorder.Three dogs had evidence of bleeding including melena (1), epistaxis (1) and petechiation (1).One dog had evidence of clot formation after a clot was formed on the tip of the dialysis catheter.Six dogs had a PLT <60 k/μL (median 35 k/μL; range, 22-49).Nine dogs had additional coagulation testing performed including PT and aPTT testing (5) and TEG (4).Of those, 8 were considered coagulopathic.Based on a combination of the testing above, 12 dogs (54%) had coagulation system involvement.
Spearman correlation analyses were used to assess the association between bilirubin and BUN and creatinine.There was no significant correlation between bilirubin and BUN (À0.08;P = .71)and creatinine (0.19, P = .40).Measures of effect could not be computed as continuous variables are presented as medians and due to the small sample size for body systems affected.Multivariate analysis (eg, logistic regression) was not possible due to small sample size and distribution of body systems affected.

| Long term outcome (≥6 months)
Of the 13 dogs that survived >30 days from discharge, 12 were alive ≥6 months from discharge and 8 were alive at the time of writing: The 8 dogs alive at time of writing were staged according to IRIS CKD staging guidelines 20 as Stage I ( 5) and Stage II (3).The median sCr was 1.3 mg/dL (range, 0.6-2.2mg/dL).Creatinine normalization occurred in 4 out of the 8 dogs (50%).Staging occurred a median of 13 months (range, 6-24 months) after discharge.
The 4 dogs not alive at time of writing were euthanized due to suspected kidney disease at 3 years (2), 2 years (1) and 7 months (1) after discharge.Serum creatinine before death was not available.The dog euthanized 7 months after discharge was newly diagnosed with metastatic mast cell neoplasia.Before euthanasia, the dog's serum creatinine was 1.0 mg/dL.None of these dogs had necropsies performed.

| DISCUSSION
Our study describes the outcomes of a cohort of dogs with severe clinical manifestations of leptospirosis that had access to RRT.
Approximately 3 /4 of the dogs survived to discharge and of those, approximately 3 /4 were alive at 6 months from discharge.Despite severe azotemia on presentation, there was no significant difference in maxBUN or maxCr between survivors and non-survivors to discharge.Non-survivors to discharge had a significantly higher number of body systems affected in comparison to survivors to discharge.
Although 75% of survivors were azotemic at discharge, kidney values continued to improve over time and normalized in 50% of long-term surviving cases.
Our study's survival rate to discharge was 73%.Other studies have found survival rates between dogs with AKI-L who did not require RRT (68%) 6 , dogs with AKI of multiple etiologies treated with RRT (60%) 23 , and dogs with AKI-L that received RRT (86%). 5The difference in survival rates might be attributed to differences in dog cohorts, fluid status and serovars.Most dogs in our study had negative prognostic factors associated with survival to discharge including overhydration (59%) and anuria or oliguria (64%). 5,6,22,23Our dog cohort had MAT titers highest to Bratislava (41%) and Gryppotyphosa (41%) serovars which differs from other studies. 55][26] The good survival rate in our AKI-L dogs with severe clinical presentation highlights RRT success when available.
Maximum blood urea nitrogen and maxCr were not associated with survival to discharge.Studies have shown that azotemia severity might be associated with short-term survival in dogs with leptospirosis, dogs with AKI and in dogs with AKI of varying etiologies requiring RRT. 6,22,23Although azotemia severity might have been underestimated due to the suppressive effect of hyperbilirubinemia on sCr, our study did not reveal an association between bilirubin and BUN or creatinine.According to our chemistry analyzer's manufacturer, creatinine suppression occurs with bilirubin levels >20 mg/dL, as seen in 6 dogs (see Data S1).Similarly to our study, other studies in RRT dogs revealed that azotemia severity is not a negative prognostic factor. 5,11e lack of association between maxBUN, maxCr and survival suggests that the severity of azotemia does not reflect the reversibility of kidney dysfunction. 22,23Leptospirosis, a potentially reversible cause of AKI, carries a more favorable prognosis in contrast to other causes of irreversible AKI where a greater degree of azotemia is associated with a worse prognosis. 22,23,27Therefore, prognosis for dogs with AKI-L that can have access to RRT should not be determined based on serum BUN and creatinine.
Non-survivors to discharge had a significantly higher number of body systems affected compared with survivors.Other studies have identified that multisystemic involvement is associated with a worse outcome. 6,23Leptospirosis has a strong ability to induce proinflammatory cytokines resulting in multisystemic disease. 28Multisystemic dysfunction adds to the complexity of case management due to the challenges associated with balancing acute liver injury, coagulation status, respiratory status and other life-threatening systemic complications in conjunction with the AKI.These challenges are illustrated by the fact that multisystemic disease was cited as a contributing factor in the decision for euthanasia in the dogs that did not survive to Comparison in number of body systems affected between survivors and non-survivors to discharge.
discharge.In human medicine, cytokine filters have been used to reduce inflammatory mediators in patients undergoing RRT with inflammatory conditions. 29,30Based on the aforementioned principle, 1 dog received treatment with a newly available cytokine filter.The effect of the cytokine filter on the dog's outcome is unclear.There are ongoing studies evaluating the utility of blood purification therapy in AKI-L dogs at our institution.
Bilirubin was higher in non-survivors compared with survivors.
Severe hyperbilirubinemia might reflect progressive disease, as RRT does not remove unconjugated bilirubin, which is plasma protein bound. 31The suspected mechanism of hyperbilirubinemia in leptospirosis is the invasion of hepatocytes' intracellular junctions by leptospires, causing bile leakage. 32However, our dog population might have had other concomitant causes of hyperbilirubinemia including hemolysis, pancreatitis, gallbladder disease, cytokine storm and its' sequela, DIC. 28,33Hyperbilirubinemia is associated with mortality in dogs and humans as it reflects manifestations of severe multisystemic disease. 6,22,31In humans, hyperbilirubinemia might also worsen kidney injury through bilirubin cast deposition. 34Our study's results are in accordance with previous studies as hyperbilirubinemia was significantly associated with survival.However, as multivariate analysis was unable to be performed, the possibility that the number of body systems affected and hyperbilirubinemia might be confounding in our study cannot be excluded.
Dogs with AKI-L receiving RRT have an overall good survival rate.
Despite 75% of dogs remaining azotemic at discharge, 50% of dogs alive at long-term follow-up had a normalization in sCr.A study on dogs managed conservatively for leptospirosis documented normalization in sCr either at discharge or at follow-up in 75% of dogs. 6Another study looking at dogs with AKI of varying etiologies found Scr normalization in 55% of dogs at discharge, and an additional 20% at follow-up. 35Although dogs remaining azotemic at and following discharge are higher than previously described, no studies have looked uniquely at normalization of kidney values in AKI-L dogs requiring RRT.The dogs in our study were likely to have sustained worse kidney injury as they met indications for RRT while the others were treated conservatively.Additionally, longterm follow-up in our study was variable and other factors including undocumented additional AKIs and concurrent diseases might have exacerbated underlying kidney dysfunction.Kidney recovery and repair might take weeks to months and continued improvement and further normalization is possible. 22,35mitations of this study included its retrospective design, small cohort of dogs and lack of standardization in diagnostics performed and long-term follow-up.Some diagnostics were performed within 24 hours of but not before RRT initiation due to emergent RRT sessions that occurred overnight.The number of body systems affected might have been underestimated as dogs were not uniformly assessed for the body systems affected.Dogs might have also had additional systems affected (ie, myocarditis, vasculitis, enteritis) although this was not included due to the difficulty in distinguishing clinical signs from uremic signs and severe electrolyte disturbances and the lack of standardized diagnostics.Additionally, hyperbilirubinemia and number of body systems affected might be correlated or confounding but multivariable analysis could not be performed.Six dogs in the study had bilirubin levels >20 mg/dL, sCr might have been suppressed in these dogs.It is unclear the degree in which concurrent diseases, age and owner finances might have negatively affected survival to discharge.
Due to a small study cohort, multivariate analysis was unable to be performed to assess for the effect of other variables including patient age, underlying CKD, fluid-status and co-morbidities on survival to discharge.There was no standardized follow-up protocol for the dogs, In conclusion, our study presents the short and long-term outcomes of a specific cohort of 22 dogs with AKI-L requiring RRT.Survival to discharge was 73% and 75% of survivors were alive at long-term follow-up (≥6 months).Seventy-five percent of dogs were azotemic at discharge and 50% of long-term survivors were azotemic on follow-up, thus longterm CKD is common in these dogs.Maximum blood urea nitrogen and maxCr were not associated with survival to discharge and should not be used to prognosticate in this group of dogs.Multiorgan involvement and hyperbilirubinemia might have negatively affected survival to discharge as these complications might not be correctable with RRT.

ACKNOWLEDGMENT
No funding was received for this study.

CONFLICT OF INTEREST DECLARATION
Mary Anna Labato consults for Boehringer Ingelheim.No other authors declare a conflict of interest.

OFF-LABEL ANTIMICROBIAL DECLARATION
Authors declare no off-label use of antimicrobials.

INSTITUTIONAL ANIMAL CARE AND USE COMMITTEE (IACUC) OR OTHER APPROVAL DECLARATION
Authors declare no IACUC or other approval was needed.

HUMAN ETHICS APPROVAL DECLARATION
Authors declare human ethics approval was not needed for this study.

ORCID
Clinical pathology evaluation performed during hospitalization was evaluated to determine the maxCr, maxBUN, maxBili and minimum platelet count (minPLT).Serum biochemistries and CBCs were performed in the clinical pathology laboratory at TCSVM, and blood smears were reviewed by a board-certified clinical pathologist if abnormalities were present.Additional testing performed during hospitalization included urine culture and sensitivity, coagulation testing before RRT.Coagulation testing included partial prothrombin time (PT) and thromboplastin time (aPTT) (Stago STart, Diagnostica Stago, Inc, Pasippany, New Jersey) and thromboelastography (TEG) (TEG 500 Thromboelastograph, Haemonetics Corp, Braintree, Massachusetts).Additional testing included heartworm and tick-borne disease testing (SNAP 4DX IDEXX), and imaging (thoracic radiographs, abdominal ultrasound, echocardiography).All imaging studies were reviewed by board-certified radiologists and echocardiograms by board-certified cardiologists.Leptospirosis testing methodologies and results were evaluated.Testing included point of care screening tests (WITNESS Lepto Rapid Test Zoetis, SNAP Lepto IDEXX), acute and convalescent MAT, and PCR on blood and urine.Microscopic agglutination test was performed according to World Health Organization and International Twenty-four dogs received treatment with RRT for management of AKI-L between January 2018 and December 2021.Two dogs without confirmed diagnosis of leptospirosis were excluded.The remaining 22 dogs were included in the study.Fourteen dogs (64%) were males (8 castrated, 6 intact) and 8 dogs (36%) were females (7 spayed, 1 intact).The median age and weight on presentation was 5.1 years (range, 0.25-11 years) and 14.0 kg (range, 6.4-29.3kg) respectively.Breeds represented included 4 mixed breeds, 3 Golden retrievers, 2 Labrador retrievers, 2 German shepherds and 1 of the following: Bulldog, Siberian husky, Pekingese, Border collie, Australian shepherd, Cairn terrier, Dachshund, Poodle, Rat terrier, West Highland White Terrier and Rhodesian Ridgeback.

F
I G U R E 1 Comparison between maxBUN, maxCr and maxBili and between survivors and non-survivors to discharge.The box represents the second and third quartiles.The horizontal line within the box represents the median.The whiskers represent the range and the circles indicate outliers.
therefore other factors including prior kidney disease, additional AKIs, medications and additional co-morbidities might have played a role in long-term patient outcome.Surviving dogs were only staged using sCr and additional biomarkers or repeat imaging were not used.Due to 4-year study duration, dogs might have been in various stages of kidney recovery when they were staged.Finally, our study describes the outcome of only a small cohort of dogs that had access to RRT.Outcomes might be different for other causes of AKI, dogs with AKI-L but not requiring RRT or dogs with AKI-L who require RRT but do not have access to RRT.
results of dogs with leptospirosis within 24 hours of first treatment with RRT.