WRITTEN PAPER Medicine Programme, Degree Project (30 hp)
Incidence, risk factors and outcome of contrast-associated acute kidney injury in the critically ill: a retrospective
bicentric cohort study with controls
Author: Ebba Eilertz Supervisor: Robert Frithiof
Date: 2018-05-22
Title: Incidence, outcome and risk factors of contrast-associated acute kidney injury in the critically ill: a retrospective bicentric cohort study with controls
Author: Ebba Eilertz
Supervisor: Robert Frithiof
Table of Contents
Populärvetenskaplig sammanfattning ... 4
Abstract ... 5
Background ... 5
Objectives ... 5
Methods ... 5
Results... 5
Conclusions ... 5
Abbreviations and acronyms ... 6
Background ... 7
History of iodinated contrast medium and acute kidney injury ... 7
Iodinated contrast medium today ... 7
Administration route... 8
Adverse reactions to iodinated contrast media ... 9
Definition of CA-AKI ... 10
Pathophysiology of CA-AKI ... 11
Epidemiology of CA-AKI... 12
Risk factors for CA-AKI ... 12
Critically Ill and AKI ... 13
Hypothesis ... 14
Question at issue ... 14
Methods ... 15
Study design ... 15
Data collection ... 15
Inclusions and exclusion criteria ... 15
Measurements and variables ... 16
Statistical analyses ... 18
Results ... 19
Baseline characteristics ... 19
Distribution of S-Creatinine at CT-scan... 21
Incidence of AKI within three days ... 22
Who gets contrast? ... 22
Which variables increase the risk for AKI? ... 22
Short term effect on S-Creatinine... 25
Long term effects on S-Creatinine ... 27
25% Increase of S-Creatinine ... 29
Differences between Uppsala and Gävle ... 30
Discussion ... 31
Main findings ... 31
Distribution of S-Creatinine ... 31
Incidence of Short term AKI ... 32
S-Creatinine in the long term ... 32
25% Increase of S-Creatinine ... 33
CA AKI in the Critically Ill ... 33
Differences between Uppsala and Gävle ... 33
Limitations of the study ... 34
Conclusions ... 35
Studies in the future... 35
Acknowledgements ... 36
References ... 37
Populärvetenskaplig sammanfattning
Kort efter den revolutionerande upptäckten av röntgen 1895 insåg många läkare att delar av kroppen inte kunde fånga upp röntgenstrålarna och följaktligen inte avbildas. Därför utvecklades bland annat joniserat kontrastmedel, en röntgentät lösning. Nu kunde organ som njurarna,
urinvägarna och levern ses, och dessutom avgränsa patologisk från normal vävnad och urskilja.
Att kunna ge kontrastmedel i samband med röntgenundersökningar är således av stor betydelse för dagens sjukvård och ger möjlighet till att i ett tidigt skede kunna upptäcka livshotande skador.
Dock finns dock en känd nackdel med kontrastmedel och det är dess skadliga effekt på njurarna och således bidragande till utvecklandet av så kallad kontrastassocierad akut njursvikt/nefropati (engelska CA-AKI, svenska KAN). Riskfaktorerna för KAN liknar de för akut njursvikt uppkommen av andra orsaker. Kritiskt sjuka patienter är ofta sviktande i flera organ och har många av dessa riskfaktorer. Det kan innebära att de är mer utsatta för KAN, men också att den additionella toxiciteten av kontrast är klinisk försumbar, ställt i relation till de diagnostiska fördelarna kontrast medför. Ifall tillförande av kontrastmedel saknar adderande betydelse hos kritiskt sjuka vad gäller utvecklandet av njursvikt vore den upptäckten av stor betydelse för framtida intensivvård. Tidigare forskning inom detta område har framförallt gjorts på patienter som fått intraarteriell kontrast i samband med kranskärlsröntgen och har därav ingen
kontrollgrupp. Av praktiska och framförallt etiska skäl föreligger en del svårigheter med dessa studier. De få studier som har gjorts med kontrollgrupp har inte kunnat påvisa en ökad risk för utvecklandet av akut njursvikt hos de patienter som fått kontrastmedel jämfört med de som inte fått det. Incidensen av KAN tycks vara densamma oberoende av givet kontrastmedel eller ej.
Denna retrospektiva studie med kontroller har studerat joniserat kontrastmedels effekt på njurfunktionen hos kritiskt sjuka och associerade riskfaktorer till akut njursvikt. Vi saknar kännedom avseende tidigare studier inom intensivvården som studerat långtidseffekten av intravenöst kontrastmedel, varför även detta har gjorts i denna studie.
Resultaten i denna studie visar inte någon signifikant skillnad gällande utvecklandet av akut
njursvikt mellan de två grupperna, vilket indikerar att kontrastmedel i sig inte är en individuell
riskfaktor i denna patientgrupp. Sannolikt är orsaken bakom utvecklandet av akut njursvikt hos
kritisk sjuka en effekt av samverkande riskfaktorer. Fler studier behövs i framtiden för att några
definitiva slutsatser ska kunna dras.
Abstract
Background
Iodinated contrast media exposure has long been known to cause damage to the kidneys. Due to multi organ failure, critically ill patients are particularly prone to develop acute kidney injury (AKI). However, whether the use of iodinated contrast medium itself is an independent risk factor of developing AKI is still controversial. Few studies assess contrast exposure in the critically ill and even fewer includes a control group, why the question still remains; is iodinated contrast media really harmful in this population?
Objectives
Primary objective was to assess whether the use of iodinated contrast medium increases the incidence of acute kidney injury in ICU patients compared to unexposed patients. Furthermore, other associated risk factors in its development was evaluated. Finally, the effect of contrast media on long term renal function was studied.
Methods
A retrospective bicentric cohort study with controls was conducted. All CT-scans ordered by the central intensive care unit at Akademiska Sjukhuset, Uppsala in 2013- 2015, and at Gävle Sjukhus, Gävle in 2015-2017 were included. Renal function was studied after administration of iodinated contrast media in the short and long term.
Results
Among patients exposed to contrast media, 32.5% developed AKI, compared to unexposed patients, where 41.0 % developed AKI. An increased S-Creatinine level at CT-scan and use of vasoactive drugs seemed to be predictive for developing AKI. Contrast media did not have any effect on long-term renal function.
Conclusions
As no difference in AKI incidence was seen in the two cohorts it may be necessary to reassess the
association of iodinated contrast media and AKI. The incidence of AKI among critically ill
patients may more likely be attributable to other causes than contrast media exposure. Contrast
media exposure does not seem to affect S-Cr in long term follow up. Data supports that an already
impaired renal function is most associated with developing AKI. However, this study cannot fully
exclude a possible effect of contrast of renal function. Larger studies are needed to confirm these
claims and bolster these findings.
Abbreviations and acronyms
AKI Acute Kidney Injury
CI-AKI Contrast Induced Nephropathy
CA-AKI Contrast Associated Acute Kidney Injury
CIN Contrast Induced Nephropathy
S-Cr Serum Creatinine
Baseline Creatinine Creatinine value 3 months prior to CT
Long term Creatinine Creatinine value 3 months post CT
KDIGO Kidney Disease Improving Global Outcomes
CT Computerized Tomography
Contrast/Contrast media Iodinated Contrast Media
Background
History of iodinated contrast medium and acute kidney injury
In 1895 the luminary Wilhelm Röntgen first detected electromagnetic radiation in a wavelength known as X-ray. Shortly thereafter scientists realized that besides the contrast between bone and soft tissue or gastro intestinal gas using this way of imaging was rather limited (1) and the use of a solution to fill potential spaces helped to depict the contrast of body organs to a far greater extent than before (2–4). Iodinated contrast media, which was one of those newly found solutions, continues to aid radiographic images and its use proves to be invaluable in the field of radiology.
Iodinated contrast media was first documented as early as the 1920s; opacification of the kidneys was seen after intravenous administration of sodium iodide in patients treated for syphilis (3,5).
Shortly thereafter it kept being used in angiography and pyelography (6). The first iodinated contrast media were derived from a triiodobenzoic acid precursor and were ionic compounds of high osmolality. By increasing the contrast between adjacent structures, the introduction of contrast media allowed for more precise imaging of anatomic structures. Contrast aids with the dentification of the interface between normal and pathologic tissue and delineation of soft tissue in contact with blood or other bodily fluid are further examples of what contrast media attributes to today (6). In some procedures involving the urinary tract and angiography like percutaneous coronary intervention contrast is not only helpful but required (7).
Despite the benefits, iodinated contrast media has a nephrotoxic effect and contributes to the development of acute kidney injury, which has been described for more than half a century (8).
There are associated risk factors that are believed to increase the risk of AKI after contrast
administration and subsequent prediction and prevention of AKI evaluation exists in the literature.
However definitive evidence of the toxicity in critically ill patients remains equivocal.
Iodinated contrast medium today
Today, several kinds of contrast media are used for different imaging procedures; however, this paper only includes iodinated contrast medias evolution and todays efforts of making it less toxic.
Iodinated contrast media is in turn used in various radiology techniques, this study has yet only
considered the risks with iodinated contrast media exposure in computerized tomography as an
examination.
The search of an ideal contrast agent with the lowest possible toxicity but still good imaging capable is still an ongoing challenge. Over the last decades, the development and improvement of iodinated contrast media has helped make remarkable significant progress in both safer and more effective compounds. Iodinated contrast media are commonly administrated intravenously and intra-arterial, but also intrathecal, enteric or via direct injection in some procedures (6).
All forms of iodinated contrast media used today are highly water-soluble carbon-based benzene rings existing as monomers with three iodine atoms attached, or as dimers with six atoms attached.
Iodinated contrast media are usually classified upon their osmolality; high-, low- and iso-osmolar (9). The first used iodinated contrast media was of high osmolality, about four times the
osmolality of blood and was further on associated with a considerably greater extent of nephrotoxicity and is therefore not used at all intravascularly today (10).
The main evolutionary issue has consequently been to make contrast agents with lower osmolality;
today low-osmolar and iso-osmolar agents are mostly used. It is acknowledged that low- and iso- osomolar contrast media are considered to cause a minor harm on renal function yet even they are potentially nephrotoxic in high risk patients and should be used in smallest possible dose(11).
Another way to classify contrast media is whether they are ionic or nonionic. Adverse reactions are more common following ionic contrast media exposure(10). It was the Swede Torsten Almén that realized the role of osmolality in iodinated contrast media and developed the first low omsolar non-ionic contras media in 1968(12).
In summary, currently four types of contrast medium are available today. High-osmolar ioninc monomers (which is scarcely used), low-osmolar ionic dimers, low osmolar non-ioninc monomers and iso-osomolar non-ionic dimers. Curiously, not a single iodinated contrast medium has been introduced all over the world since 1995 (13).
In patients with normal kidney function a majority of the contrast medium is excreted in 24 hours.
Patients with impaired renal function the elimination half-life increase to about 40 hours.
Administration route
Way of administration is according to some also believed to affect the risk of CA-AKI, and some
studies claim that interatrial administration is related to a higher risk compared to intravenous
administration (14). The Canadian association of radiologists says that the risk of CA-AKI after
intra-arterial administration of contrast media appears to be at least twice as high compared to intravenous administration (15), while some more recent studies doubt that and question the pathophysiology behind that statement (7,16). Nyman et al. mention several reasons behind that belief in their meta-analysis from 2012; for instance, all data regarding this may be bias as high- risk patients mostly are excluded from CT-scans and receiving intravenous contrast media, while high-risk patients cannot be excluded from life-saving coronary procedures. One study did compare the incidence of AKI following intravenous administration and intra-arterial
administration of iodixanol (17). Karlsberg et al were in their study able to show that the rates of CA-AKI were not statistically different between intra venous and intra-arterial infusion in the same population (17). Further on, there is a lack of comparable studies evaluating the risk of CA- AKI between intra-arterial and intravenous administration, with matched risk factors and doses of contrast media (16).
Adverse reactions to iodinated contrast media
As previously mentioned, regardless of today’s effort of making it less toxic, iodinated contrast media has a backside and can cause different kinds of reactions after administration. The
reactions are classified as either general adverse reactions or renal adverse reaction (18). The non- renal reactions are divided into acute, late and very late. Acute reactions are mostly allergy-like or hypersensitivity reactions and includes everything from mild urticaria and itching to hypotensive shock and cardiac arrest. Patients can also experience nausea, anxiety or other chemotoxic effects.
Late adverse reactions, which occurs one hour to one week after the contrast medium is injected, mostly includes skin reactions while very late adverse reaction, occurring more than one week after injection, are thyrotoxicosis or nephrogenic systemic fibrosis. The latter is mostly following gadolinium-based contrast while thyrotoxicosis is more common after iodinated contrast media injection. Hence, at risk are patients with untreated Graves’ disease and patients with reduced renal function or patients on dialysis(18).
Above mentioned general adverse reactions are important to be aware of before administrating
contrast media to a patient. Although, this paper focus on the renal adverse reactions following
iodinated contrast media exposure, which exceedingly should be avoided. European Society of
Urogenital Radiology refers to this event as post-contrast acute kidney injury, PC-AKI, and define
it as we, among most other recent studies, do; an increase in serum creatinine by ³26.5µmol/l or
1.5 times baseline creatinine, within 48-72 hours after administration(19). Patients at risk is the
ones with eGFR less than 45 ml/min/1.73 m
2, if contrast medium is administrated intra-arterial with first pass renal exposure, or in ICU patients; which in this paper is the case. Regarding intravenous administration and intra-arterial administration with second pass renal exposure, they suggest patient with eGFR less than 30 ml/min/1.73 m
2,should refrain contrast media, i.e.; they claim intravenous administration to be less risk full (19).
Definition of CA-AKI
Contrast associated acute kidney injury is a rather new term, and the condition when patients develop kidney injury followed by contrast media exposure has previously mostly been referred to as CIN (contrast induced nephropathy) or CI-AKI (contrast induced acute kidney injury).
Contrast induced nephropathy refers to an event occurring after contrast media infusion not attributable to other causes, while using the word association is a wider term and is, in my opinion, more correct to use while debating about whether contrast media really an independent risk factor is. A scarce amount of studies uses that term, yet they, understandable, are the ones that have been done in an ICU setting; which all challenge the evidence for a causal association
between iodinated contrast media and AKI (20–26).
There are several ways to define CA-AKI, and the use of the broader definition yields a greater incidence of AKI following contrast media exposure (20). Previously, the most common definition used in literature was an increase of S-Cr with 44.2µmol/l (0.5mg/dl) and/or an increase of S-Cr with 25% within three days after contrast exposure (27–30).
Nowadays though, KDIGO guidelines, in accordance with the RIFLE and AKIN criteria (31),
recommend using the general definition of AKI, which is shown in table 2 below. Therefore, CA-
AKI was in this study defined according to KDIGO Guidelines definition of AKI, whereas it is
defined as any of the following; increase in SCr by ³0.3mg/dl (³26.5µmol/l) within 48 hours; or a
minimum of 50% increase of baseline, which is known or presumed to have occurred within the
seven prior days, or urine volume <0.5 ml/kg/h for 6 hours (32). Furthermore, AKI is staged for
severity according to as the table below shows.
Table 1 - AKI definition. AKI was in this study defined as the table below shows.
Pathophysiology of CA-AKI
Acute kidney injury is clinically divided into three etiologies; prerenal, renal and postrenal(33).
One of many important functions of the kidneys is the filtration and excretion of nitrogenous waste products form the blood. Acute kidney injury is a rather common condition in hospitalized patients and associated with high mortality rates(34).
Iodinated contrast media exposure is a well-known cause to acute kidney injury, but the mechanisms behind it are yet not completely understood. Its direct tubular damage, intrarenal vasoconstriction with consequential medullary hypoxia and generation of reactive oxygen species are the predominant factors in the complex poorly understood mechanism (35).
Several animal experiments show that CA-AKI is accompanied by an increased production of ROS (36,37). One study made both in vitro and in vivo concludes that CM-induced tubular renal cells apoptosis represents a key mechanism of CA-AKI(37).
AKI stage Serum creatinine Urine output
1 1.5-1.9 times baseline or
³0.3mg/dl (³26.5µmol/l) increase
<0.5 ml/kg/h for 6-12 hours
2 2.0-2.9 times baseline <0.5 ml/kg/h for ³12 hours
3 3.0 times baseline or
³ 4.0 mg/dl (³ 353.6 µmol/l) in serum creatinine or
Initation of renal replacement therapy or in patients <18 years, decrease in eGFR to
<35ml/min per 1.73m
2<0.3 ml/kg/h for ³24 hours or
Anuria for ³12 hours
Epidemiology of CA-AKI
CA-AKI has in many studies been described as the third biggest reason to development of acute kidney injury in hospitalized patients(38–40), after decreased renal function and nephrotoxic medications (32). As previously mentioned the incidence of CA-AKI differs a lot depending on which definition the study has used; using a broader definition yields a greater incidence. Different patient population, dose of contrast and baseline risk might also affect the incidence(32). The epidemiological studies of CA-AKI have long been hampered by above mentioned reasons, and further on by the fluctuation of S-Cr and creatinine clearance during hospital stay and ICU- admission(41).
McDonald et al describe in their meta-analysis 2013 how the incidence of CA-AKI varies between 1-30%, and also that the risk ratio of mortality after CA-AKI ranges widely(42). Hoste et al show in their retrospective single-center study that CA-AKI occurs in on out of six (16.3%) critically ill patients who underwent a radiography enhanced by contrast media, however this study does not include a control group(25). Actually, Ehrmann et al showed in their systematic review and meta- analysis from 2017 that 95% of the studies they reviewed were lacking a control group. This demonstrates the need of future studies including a control group to evaluate the actual incidence of CA-AKI(22).
Risk factors for CA-AKI
Many of the risk factors of CA-AKI are similar to the ones of AKI. The Canadian association of Radiologists explain in their Guidelines from 2012 that the major risk factor predicting contrast induced nephropathy is pre-existing renal failure(15). KDIGO Guidelines, among most studies within this area, agrees and says that the most important risk factor to screen for before
intravascular administration of iodinated contrast media is an already impaired renal function(21,27,32,43,44). In those patients, the risk is even more elevated and of clinical importance if combined with diabetes(45). Besides chronic kidney disease (CKD), other risk factors of developing CA-AKI include diabetes mellitus, chronic heart failure, advanced age, volume depletion, hemodynamic instability, use of additional nephrotoxic drugs, and large volume or high osmolality of the contrast agent(32,46).
However, all studies mentioned above have been done on patients undergoing cardiovascular
procedures such as cardiac catheterization and percutaneous interventions.
Recent studies in ICU settings have shown the same risk factors of CA-AKI in this patient group too, and further on focus more on the fact that many of those conditions are common in the critically ill; consequently, question the real impact of iodinated contrast media (15,16,18). Those risk factors are also similar to the ones of developing AKI in general, regardless underlying cause (47). Sequential Organ Failure Assessment (SOFA) score and number of nephrotoxic agents (other than contrast media) were independent risk factors in one study recently made in an ICU-setting (20).
Critically Ill and AKI
Critically ill patients often have several risk factors of developing acute kidney injury, regardless iodinated contrast media exposure or not. Consequently, the question whether the administration of contrast media really is harmful in these patients has arisen(20–25,48). The risk factors for AKI in critically ill patients are similar to the ones of CA-AKI(49–51). The incidence of AKI in
critically ill patients varies between 1-70% depending on criteria used(51). AKI is in most cases associated with an increased need for renal replacement therapy and increased mortality
rate(51,52).
Septic shock, which is a common condition in critically ill patients, has in many studies been associated with a high risk of developing AKI; Bagshaw et al demonstrate, in a large retrospective multicenter study, that as many as 64,4% of the patients with septic shock developed early AKI, i.e., within 24 hours after onset of hypotension(53). This is one of many things that shows how vulnerable critically ill patients are and prone to develop AKI, possibly including CA-AKI, considering latest uncertainties about it being a real thing or not(23). Further causes to systematic inflammatory response such as trauma is also common in ICUs; with associated biological processes, which contributes to organ injury(54). Moreover, the use of other nephrotoxic drugs other than contrast media is also common in critically ill patients(55).
Having said that, on one hand, critically ill patients may be of a greater risk than other patients developing CA-AKI through a decline in an already ongoing kidney injury(20). On the other hand, this makes one wonder what the real attributable risk for renal dysfunction from iodinated contrast media is.
Further on, critically ill patients pose a great risk of life-threatening illness and conditions which
can be confirmed or excluded by the help of radiological imaging enhanced by iodinated contrast
media. Physicians often have to decide whether they should proceed with a radiological study with iodinated contrast media to confirm or exclude a life-threatening condition, that may be reversible or treatable (21). This stresses the need to evaluate the benefit-harm ratio of iodinated contrast media on these patients, which can be made by assessing whether iodinated contrast media adds any further risk of developing acute kidney injury.
Most of the studies made about iodinated contrast media exposure and its believed causation to development of acute kidney injury have been made in coronary angiography and percutaneous intervention, hence the lack of a control group. In patients undergoing CT-scan with intravenously administrated contrast media there is a possibility to include a control group though, why we did this retrospective study with controls in two ICU-setting in an effort to evaluate the real incidence and risk of acute kidney injury attributable to iodinated contrast media exposure. As far as we know no study in an ICU setting has evaluated the effects of iodinated contrast media in the long term, which this study also did by measuring S-Cr 3 months after contrast exposure.
Hypothesis
Iodinated contrast media exposure does not affect the development of Acute Kidney Injury in the critically ill within three days after exposure. Neither does it have any effect on renal function/
serum creatinine levels three months after the CT-scan.
Question at issue
Does iodinated contrast medium affect the development of AKI in critically ill patients that have underwent a CT-scan?
Does iodinated contrast media have any long-term effects on S-Creatinine changes?
Are there any other important risk factors for the development of AKI within three days after a
CT-scan?
Methods
Study design
This study was a bicentric retrospective cohort study, with controls, in the central intensive care unit (CIVA) at Akademiska Sjukhuset in Uppsala, Sweden and in the central intensive care unit at Gävle Sjukhus in Gävle, Sweden, made by another medical student. Medical, surgical and trauma patients are committed to these units which gives the study a wide range of critically ill patients.
The study has been approved by the Regional Ethics Committee in Uppsala. The approval came 2017-08-09 and the Dnr is 2017/168.
Data collection
The data collection consisted of gathering multiple variables from several medical record systems used in Uppsala and Gävle. All the CT-scans committed by CIVA during 2013, 2014 and 2015 were registered at the Radiology Department at Akademiska Sjukhuset and listed with all needed information about the examination; if the patient had received contrast media, and in those cases dose and type of contrast media were also noted. A total of 1007 examinations in 729 patients were found. Most variables from the patients’ medical history was taken from Cosmic, which is the main medical record system used at Akademiska Sjukhuset. Data specific to intensive care unit patients, like SAPSIII score etc. was taken from PasIVA, which is a system used in intensive care units. Kovis, a platform for scanned material, has also contributed with some data. Some of the newest data have been taken from Metavision, which is a new system used for anesthesia documenting during surgery and in intensive care units. In Gävle, the similar has been done but instead from the medical record system Melior, and also PasIVA. Access to all those medical records was approved of Suzanne Odemark Wernerman, head of department at the intensive care unit in Uppsala.
Inclusions and exclusion criteria
All CT-scans committed from the Central Intensive Care Unit in Uppsala for three years, 2013- 2015, and in Gävle for three years, 2015-2017, were included. Patients were excluded in cases of age<16 years, renal replacement therapy pre-CT-scan, lack of serum creatinine measured day of examination or the three following days, and ultimately length of ICU-stay less than 48 hours.
Patients undergoing more than one examination during their ICU-stay were only included once.
Measurements and variables
Patient characteristics were noted as follows; demographic data such as gender and age were collected from the patients’ medical records, and clinical data included medical history, main diagnosis at ICU admission, acute failures during ICU admission, AKI development, S-Cr values.
In terms of medical history, it was recorded if the patients had any of following conditions;
hypertension, chronic heart failure, liver failure, diabetes mellitus and chronic kidney disease.
Main diagnosis of ICU admission was divided into circulatory failure, acute respiratory failure, cardiac arrest, infection, trauma and other. Other included bowel ischemia, intoxication, liver failure, ketoacidosis, multi organ failure, esophageal rupture, hyponatremia, oliguria, kidney rupture, post op observation, acute kidney injury and aortic dissection type B.
Acute conditions during ICU admission such as acute heart failure and acute liver failure were recorded. Acute heart failure was referred to as rapid onset or worsening of symptoms and signs of heart failure (56,57). Medical records during ICU admission mentioning any of these
(breathlessness, elevated jugular venous pressure, pulmonary crackles or peripheral edema) were noted, and it was noted if the patients had done an echocardiography or other imaging procedure during their ICU stay confirming an acute heart failure. In this study it was defined according to European Association for the Study of the Guidelines; two- to three times elevation of
transaminases (as a marker of liver damage) associated with impaired liver function, i.e., jaundice and coagulopathy, in a patient without a chronic liver disease (58).
S-Cr values were recorded (if possible) at six different times; a minimum of three months prior to
CT-scan, day of ICU admission, day of CT-scan, daily following 72 hours and finally a long-term
value, i.e. a minimum of three months post procedure. If several values were found close to each
other a mean value was calculated, both regarding baseline and long-term. If a baseline S-Cr
could not be found a formula recommended by KDIGO r was used to estimate this; however, this
study did not, unlike the recommendations from KDIGO, take ethnicity into account but age and
gender (32). Urine output is measured hourly in most ICU settings and was in this study recorded
a bit differently between Uppsala and Gävle, 12 h respective 24 h pre and post administration of
contrast media. In both settings daily urine output was recorded and used together with S-Cr
measures to calculate stage of AKI in accordance the KDIGO Guidelines definition, day of
inclusion and following three days.
Characteristics at inclusion were further on noted; mechanical ventilation (invasive or non-
invasive), PaO2/FIO2 (mmHg), mean arterial pressure (mmHg) at the time of CT-scan and lowest during ICU admission, highest arterial lactate (mmol/L), lowest hemoglobin concentration (g/dL).
Administrated drugs recorded at inclusions were nephrotoxic drugs (see table 2), vasodilators (Noradrenaline, Adrenaline, Dobutamine, Dopamin, Efedrin, Atropin,), any antibiotics and diuretics. Patients were considered receiving nephrotoxic drugs, vasodilators and antibiotics if administrated 24 hours pre or post procedure, and diuretics 48 hours post procedure.
Contrast exposure was specified in dose and type of contrast; most patients received the non-ionic low osmolar monomer Omnipaque 350, contacting the active substance iohexol. 350 is the iodid concentration in mg/mol and this agent has an osmolality of 730 mOsm/kg H
2O. Two patients received Xenetix350, ioibitridol, and two received Iomeron400, jomeprol, both non-ionic low osmolar monomers.
Administration of N-Acetylcysteine before contrast inclusion were noted; however none specific preventive strategies for CA-AKI are done at these two ICU settings considering all critically ill patients already are optimized in fluid balance regardless contrast media exposure. SAPSIII (Simplified Acute Physiology Score III) score, which is a predictive score system for mortality in the critically ill (59,60), was recorded from the ICU-specific system PasIVA. The score predicts a patient’s mortality during ICU stay by using various parameters; age, gender, reason to ICU admission, medical history, previous surgery, vitals, lab results etc.
Table 2 - Nephrotoxic drugs. The drugs listed below were the ones considered nephrotoxic in this study. It was noted if a patient received any at all, and if they were; how many and which ones. 232 of 367(data could not be found in all patients) patients (63,2%) received nephrotoxic drugs.
Aminoglycosides Mannitol
Vancomycin Cyclosporines
Amphotericin B Tacrolimus
Aciclovir Hydroxyl-ethyl-starch (HES)
Angiotensin II Converting Enzyme –
Inhibitors Non-steroidal anti-inflammatory drugs
(NSAID)
Angiotensin II Receptor Blockers Colistin
Diuretics Gadolinium
Statistical analyses
The statistician Johan Westerbergh has done all statistical analysis at Uppsala Clinical Research Center. All the analysis has been made in R version 3.3.2. I the tables illustrating patient characteristics discrete variables were recorded as percentage of total patients within each cohort and continuous variables were described with the median value ad separated into quartiles.
Wilcoxon test was used for the univariate analysis of the quantitative variables and Fisher’s exact test was used for the qualitative variables. For S-Creatinine at CT an analysis of variables(anova) was performed, to evaluate if there was any interaction between the group exposed to contrast and the unexposed group, and between the two hospitals. The analysis of how contrast is given was performed using logistic regression, first with the model Contrast= Crea.CT + Age + SAPS.III + Vasodil.drugs + Nephrotoxic.drugs + Ventilation + hosp. In the same way the questions which variables increases the risk for AKI was answered by using multivariate logistic analysis, first with base adjustment just locking at AKI = ContrastCT + Crea.CT, and additionaly with full adjustment with 9 variables; AKI = Contrast.CT + Crea.CT + Age + SAPS.III + Vasodil.drugs + AKI.grad.CT + Nephrotoxic.drugs + Ventilation + hosp. Finally a small adjustment were made looking at only 4 variables; AKI = Contrast.CT + Crea.CT + Vasodil.drugs + Contrast.CT:Crea.CT. The short-term effect on S-Cr was analyzed using linear regressions with the same set of variables used for logistic regression (full adjustment). The model was fitted to the ln-difference between day 1-3 and S-Cr at CT (ln (short crea/crea at CT)). Finally, the same analysis as for short term was done for long term but the ln-difference between long term S-Cr and baseline S-Cr (ln (long crea/baseline)). P-values
<0,05 were considered significant in all statistical analysis.
Results
Baseline characteristics
Table 3 – Baseline characteristics between cohort undergoing CT with contrast and the cohort undergoing CT without contrast. Discrete variables were recorded as percentage of total patients within each cohort, and continuous variables were described with the median value and separated into quartiles. Wilcoxon test was used for the univariate analysis of the quantitative variables and Fisher’s exact test was used for the qualitative variables. “AKI within three days” refers to three days from CT-scan.
2 RESULTS 2.1 Baseline characteristics
Table 2. Baseline table constrast. Open the table in an Excel file by clicking .
Contrast at CT
N No Yes Combined P -value
N = 213 N = 168 N = 381
Hospital
Hospital: G¨avle 381 75 (35.2%) 53 (31.5%) 128 (33.6%) 0.511
Demography
Age 380 67.0 (51.8 – 72.0) 65.0 (48.8 – 73.0) 66.0 (51.0 – 73.0) 0.502
Gender: Female 381 66 (31.0%) 73 (43.5%) 139 (36.5%) 0.0141
Acute kidney injury
AKI within 3 days 376 86 (41.0%) 54 (32.5%) 140 (37.2%) 0.111
AKI-grade at CT: 0 373 143 (68.8%) 136 (82.4%) 279 (74.8%) < 0.0011
1 20 (9.6%) 17 (10.3%) 37 (9.9%)
2 25 (12.0%) 10 (6.1%) 35 (9.4%)
3 20 (9.6%) 2 (1.2%) 22 (5.9%)
Contrast
Dose of contrast (ml/kg) 377 0.0 (0.0 – 0.0) 1.0 (0.8 – 1.2) 0.0 (0.0 – 0.9) < 0.0012 Cause of admission
Cause of admission: Circulatory failure 288 11 (6.8%) 3 (2.4%) 14 (4.9%) —
Heart arrest 17 (10.6%) 8 (6.3%) 25 (8.7%)
Infection 37 (23.0%) 29 (22.8%) 66 (22.9%)
Respiratory failure 39 (24.2%) 51 (40.2%) 90 (31.2%)
Trauma 33 (20.5%) 22 (17.3%) 55 (19.1%)
Other 24 (14.9%) 14 (11.0%) 38 (13.2%)
Medical history
Hypertension 380 86 (40.4%) 56 (33.5%) 142 (37.4%) 0.201
Congestive of heart failure 336 33 (17.6%) 15 (10.1%) 48 (14.3%) 0.0601
Liver failure 337 12 (6.3%) 8 (5.4%) 20 (5.9%) 0.821
Diabetes Mellitus 380 39 (18.3%) 28 (16.8%) 67 (17.6%) 0.791
Chronic Kidney Disease 336 20 (10.5%) 3 (2.1%) 23 (6.8%) 0.0021
Medical events and clinical markers3
SAPS III 381 64.0 (54.0 – 74.0) 60.0 (50.0 – 68.0) 62.0 (52.0 – 72.0) 0.0112
Acute heart failure 379 33 (15.6%) 12 (7.2%) 45 (11.9%) 0.0161
Acute liver failure 379 20 (9.4%) 10 (6.0%) 30 (7.9%) 0.251
Highest lactate level 323 2.8 (1.7 – 4.7) 2.4 (1.7 – 4.2) 2.7 (1.7 – 4.4) 0.292 Lowest Hb 327 88.0 (80.0 – 101.0) 85.0 (80.0 – 96.2) 87.0 (80.0 – 99.0) 0.482 Highest mean arterial pressure 376 55.0 (50.0 – 60.0) 55.0 (50.0 – 60.0) 55.0 (50.0 – 60.0) 0.272 PaO2/FiO2 at CT 366 27.0 (20.0 – 39.0) 27.2 (19.9 – 38.1) 27.0 (20.0 – 38.7) 0.802 Creatinine
Baseline Creatinine 379 89.7 (72.4 – 99.9) 78.0 (70.0 – 92.2) 84.0 (70.3 – 97.0) < 0.0012 Creatinine at CT 378 91.0 (66.0 – 171.5) 76.0 (60.0 – 99.5) 84.0 (62.0 – 127.8) < 0.0012 Long term creatinine 191 75.0 (64.0 – 91.5) 68.5 (55.0 – 82.9) 71.0 (59.5 – 89.5) 0.0202
25 % increase in creatinine 371 30 (14.6%) 29 (17.6%) 59 (15.9%) 0.481
Medication
Diuretics Post CT 376 107 (51.2%) 90 (53.9%) 197 (52.4%) 0.611
Vasoactive drugs 378 143 (68.1%) 131 (78.0%) 274 (72.5%) 0.0371
Number of vasoactive drugs: 0 311 22 (13.3%) 15 (10.3%) 37 (11.9%) 0.701
1 112 (67.9%) 107 (73.3%) 219 (70.4%)
2 24 (14.5%) 18 (12.3%) 42 (13.5%)
3 7 (4.2%) 5 (3.4%) 12 (3.9%)
4 0 (0.0%) 1 (0.7%) 1 (0.3%)
Nephrotoxic drugs 371 130 (63.4%) 104 (62.7%) 234 (63.1%) 0.911
Number of nephrotoxic drugs: 0 301 42 (24.3%) 24 (18.8%) 66 (21.9%) 0.351
1 106 (61.3%) 81 (63.3%) 187 (62.1%)
2 23 (13.3%) 23 (18.0%) 46 (15.3%)
3 2 (1.2%) 0 (0.0%) 2 (0.7%)
4 0 (0.0%) 0 (0.0%) 0 (0.0%)
Ventilation
Mechanical ventilation 381 186 (87.3%) 154 (91.7%) 340 (89.2%) 0.191
Type of ventilation: Non-invasive 145 59 (68.6%) 45 (76.3%) 104 (71.7%) 0.351 m (a – b) represents median (Q1– Q3).
n (p%) represent frequency (percentage). Percentages computed by group.
Tests used:1Fisher’s exact test;2Wilcoxon test.
3Taken at intensive care unit.
Johan Westerbergh
Uppsala Clinical Research Center 3(15) May 31, 2018
1007 examinations in 739 patients were recorded. In Uppsala it was a total of 686 examinations in 524 patients and in Gävle 319 examinations in 215 patients. After exclusions 585 CT-scans remained. After including every patient only once 381 examinations were analyzed; 128 from Gävle and 253 from Uppsala. Of these 381 examinations 168 (44.0%) received contrast, 53 (31.5%) of the patients in Gävle and 113 (68.5%) in Uppsala. Existing data of serum creatinine values three months after the radiological procedure were found in 191 of 381 patients.
Table 3 shows discrete variables with the percentage of patients within each cohort, and
continuous variables were described with median value separated into quartiles. Data compares
patients that received contrast media at CT-scan to the ones who did not. The study sample
included 382 patients. However, seven patients did not have data regarding creatinine or urine
output and were not included in the AKI incidence analysis. In Uppsala patients missing he table
showing AKI incidence and distribution has 376 patients; data regarding creatinine or urine output
is lacking for 5 patients. In Uppsala these patients with missing renal function and urine output
data were excluded but were included in Gävle, hence the confusing total number of examinations
analyzed.
Distribution of S-Creatinine at CT-scan
Creatinine (µmol/L)
Figure 1 - Distribution of creatinine at CT-scan. The figure shows the distribution of creatinine between patients that received contrast at CT and those who did not, comparing Uppsala to Gävle. One dot shows one patient’s creatinine value in µmol/L and the lines show the distribution of all the values. Red observations received contrast at C and blue did not.
Figure 1 above shows the distribution of S-Cr level between contrast exposure cohort and control cohort. S-Cr dots show the distributions between individuals exposed to contrast and individuals unexposed.
Table 4 – Distribution of creatinine at CT-scan test of significance. Test of significance in S-Cr level differences between contrast exposed and unexposed patients, between the two hospitals and the interaction.
Analysis of variance (ANOVA) was performed fort his.
Estimate Std.Error t-value p- value
(Intercept) 4.68 0.05 94 <0.001
hospGävle -0.05 0.083 -0.64 0.524
Contrast.CTYes -0.31 0.074 -4.17 <0.001
hospGävle:Contrast.CTYes -0.02 0.126 -0.19 0.846
Table 4 shows a significance difference in S-Cr value day of CT-scan between the ones that
received contras media and the ones that did not.
Incidence of AKI within three days
Using the KDIGO definition(32), acute kidney injury within three days after contrast media exposure occurred in 140 of 376 patients (37.2%). In the group exposed to contrast media, 54 of 168 (32.5 %) patients developed AKI within three days versus 86 of 213 (41.0%) in the unexposed group.
Who gets contrast?
2a) 2b)
Figure 2 - Analysis of variance. Showing relationship between variables and contrast at CT (2a), test of
significance. 2b shows the same relationship but with odd rations with confidence intervals.
Figure 2 shows the association of contrast exposure with CT to patient related factors. Patients with higher S-Cr were less likely to receive contrast, and patients on vasodilators were more likely to receive contrast media at time of CT-scan, which both were significant. No other variables were significant when receiving contrast media at time of CT-scan.
Which variables increase the risk for AKI?
To study potential risk factors and variables associated with AKI, logistic regression models were used for relationship between variables and AKI. Base adjustment for patients with contrast exposure and S-Cr at CT as variables. Full adjustment (nine variables) and small adjustment (four variables) was also performed.
3a) 3b)
Figure 3 - Base adjustment. Relationships between variables and AKI, odds ration with confidence intervals and test of significance. AKI =Contrast.CT +Crea.CT. P-values <0.05 were significant.
4a) 4b)
Figure 4 - Small adjustment. Relationships between variables and AKI, odds ration with confidence intervals and test of significance. AKI = Contrast.CT + Crea.CT + Vasoaktiva.lm + Contrast.CT:Crea.CT. P-values <0.05 were significant. Figure 4b shows the variables ranked after most impact on AKI. Y-axis shows x2-value from Walds test, next to a consistent p-value. X-axis shows x2-value minus degree of freedom(df).
5a) 5b)
Figure 5 - Full adjustment. Relationships between variables and AKI, odds ration with confidence intervals and test of significance. AKI = Contrast.CT + Crea.CT + Age + SAPS.III + Vasoaktiva.lm + AKI.grad.CT + Nefrotoxiska.lm + Ventilation + hosp. P-values <0.05 were significant. 5b shows the variables ranked after most impact on AKI. Y-axis shows x2-value from Walds test, next to a consistent p-value. X-axis shows x2-value minus degree of freedom(df).
Figure 3 demonstrates the relationship between receiving contrast and AKI and S-Cr at CT -and AKI. S-Cr has a positive significant effect on AKI, whereas contrast does not. In Figure 4
Chi.Square d.f. P Contrast.CT 2.448 1 0.118 Crea.CT 65.977 1 <0.001 TOTAL 66.580 2 <0.001
0.50 1.00 2.00 4.00 7.00 Crea.CT − 126:62
Contrast.CT − Yes:No
vasoactive drugs is added as a variable, and also the interaction of contrast and S-Cr at CT;
evaluating if the risk of AKI would increase more for patient with higher S-Cr at CT if they received contrast media. The figures show that both the interaction of S-Cr and contrast and contrast itself is insignificant, however, vasoactive drugs and S-Cr at CT-scan are still significant and associated with a higher incidence of AKI. Table 5 illustrates the same thing using a different model with more variables. Interestingly, there is a significant difference between the hospitals with a higher risk of AKI in Gävle, considering the variables in that model. AKI present at the time CT is performed was also significant and associated with a higher risk for AKI three days following the CT-scan, understandably.
6a) Contrast at CT 6b) Vasoactive drugs
As a final analysis in studying variables most associated with AKI an estimated rate of AKI using continuous level of creatinine was calculated. Figure 6a shows the rate of AKI for contrast at CT where no significant differences can be seen. The upper panel shows group with vasoactive drugs and the lower without. Figure 7a illustrates the same but for patients with vasoactive drugs and patients without the usage of vasoactive drugs. Upper panel shows patients exposed to contrast media and lower shows patients unexposed to contrast media. Unlike figure 6a, in figure 6b a significant difference can be seen; both in patients exposed and unexposed to contrast media.
Figure 6 - Estimated rate of AKI using continuous level of creatinine for contrast at CT (yes/no) and usage of vasoactive drugs (yes/no), with 95% confidence intervals. Figure 6a shows the creatinine effect on rate of AKI for patients exposed to contrast (red lines) and patients unexposed (blue lines); upper panel shows with vasoactive drugs and lower without. Figure 6b shows the creatinine effect on rate of AKI for patients with vasoactive drugs (red line) and patients without vasoactive drugs (blue line);
upper panel shows patients exposed to contrast media and lower unexposed patients. The red and blue areas show the 95%
confidence intervals for the lines.
Short term effect on S-Creatinine
7a) 7b)
7c)
Creatinine at CT (µmol/L)
Figure 7 - Short term effect on S-Creatinine. Difference between (lndiff) between S-Cr at CT and 1-3 days later, (ln(Creatinine day x/creatinine at CT). Red line shows patients exposed to contrast and blue line shows patients not exposed. Each dot shows one individual value and the lines illustrate linear regressions lines with S- creatinine at CT as variable for the two just mentioned groups. Figure 7a shows the result for day 1, 7b for day 2 and 7c for day 3. The gradient of the lines shows the effect of S-Cr at CT-scan, a higher value generates a greater decrease.
Figure 7 above show the differences between S-Cr at CT-scan and day 1, day 2 and day 3. The
difference is shown in logarithm values, i.e. in proportion of each other, instead of their absolute
number. A greater decrease can be seen the higher S-Cr value a patient has. In day 1 there is a
difference between the two groups and it looks like patients unexposed to contrast had a decline in
S-Cr whereas the exposed group did not. That could be explained by the fact that the unexposed
group is more likely to have higher S-Cr at CT, and therefore could be more likely to have a
decrease the day after. However, this was not significant, as shown in table 5 below.
Table 5 - Short term effect in S-Cr. Testing if the change of S-Cr between CT-scan and day 1, 2 and 3 can be explained by any of the following variables. The results are showing the linear regression models with the logarithm difference in S-Cr for each day compared to S- Cr at CT. P-values lower then 0,05 was significant.
S-Cr levels at CT is significant in day 2 and 3, and nearly in day 1. At day three S-Cr at CT is strongly significant, with a negative estimation, meaning higher S- Cr is often lower following days after procedure, especially three days after.
Besides S-Cr at CT, AKI at CT and nephrotoxic drugs were significant. AKI at CT is negative meaning the ones with AKI at CT has a higher chance of
decrease. Nephrotoxic drugs are positive which means that these patients in general experience an increase in S-Cr.
Most interesting in this analysis is yet the non-existing effect of contrast exposure at CT, with insignificant p- values all the three days.
Estimate Std.Error t- value p-
value Day 1
(Intercept)
-0.1376
0.0653 -2.11
0.036 Contrast.CTYes -0.0692 0.0466 -1.48 0.139 Crea.CT -0.0005 0.0002 -2.77 0.006
Age 0.0001 0.0007 0.09 0.928
SAPS.III 0.0010 0.0009 1.12 0.263
Vasoaktiva.lmYes 0.0405 0.0285 1.42 0.157 AKI.CT1 -0.0192 0.0370 -0.52 0.604 Nefrotoxiska.lmYes 0.0251 0.0254 0.99 0.322 VentilationYes 0.0522 0.0392 1.33 0.184 hospGävle 0.0601 0.0258 2.33 0.020 Contrast.CTYes:Crea.CT 0.0006 0.0004 1.43 0.154 Day 2
(Intercept)
-0.3047
0.0891 -3.42
<0.001 Contrast.CTYes 0.0146 0.0627 0.23 0.816 Crea.CT -0.0006 0.0002 -2.89 0.004
Age 0.0013 0.0010 1.30 0.195
SAPS.III 0.0012 0.0012 1.00 0.317 Vasoaktiva.lmYes 0.0214 0.0384 0.56 0.578 AKI.CT1 -0.0465 0.0491 -0.95 0.344 Nefrotoxiska.lmYes 0.0805 0.0342 2.36 0.019 VentilationYes 0.1149 0.0534 2.15 0.032 hospG¨avle -0.0033 0.0347 -0.10 0.924 Contrast.CTYes:Crea.CT 0.0000 0.0006 0.07 0.945 Day 3
(Intercept)
−0.3548
0.1068 -3.32
0.001 Contrast.CTYes 0.0097 0.0751 0.13 0.897 Crea.CT −0.0009 0.0003 -3.50 <0.001
Age 0.0005 0.0012 0.43 0.669
SAPS.III 0.0024 0.0015 1.60 0.110
Vasoaktiva.lmYes 0.0092 0.0448 0.20 0.838 AKI.CT1 −0.1109 0.0576 -1.93 0.055 Nefrotoxiska.lmYes 0.1130 0.0404 2.79 0.006 VentilationYes 0.1271 0.0646 1.97 0.050 hospG¨avle −0.0457 0.0410 -1.11 0.266 Contrast.CTYes:Crea.CT 0.0002 0.0007 0.34 0.738
Long term effects on S-Creatinine
Table 6 - Effect of contrast media on long term S-Cr. The results show two linear regression models with difference between baseline S-Cr and long-term S-Cr. P-value <0.05 was significant.
Estimate Std.Error t-value p-value (Intercept) −0.4498 0.1459 -3.08 0.002 Contrast.CTYes 0.0685 0.1143 0.60 0.550
Crea.CT 0.0005 0.0003 1.69 0.093
Age 0.0044 0.0016 2.75 0.007
SAPS.III 0.0008 0.0022 0.37 0.709 Vasoaktiva.lmYes −0.0282 0.0603 -0.47 0.640
AKI.CT1 0.1823 0.0803 2.27 0.024
Nefrotoxiska.lmYes −0.0455 0.0550 -0.83 0.409 VentilationYes −0.0310 0.0762 -0.41 0.684 hospGävle −0.0106 0.0590 -0.18 0.858 Contrast.CTYes:Crea.CT −0.0004 0.0012 -0.35 0.728
The only significant variable on difference between baseline S-Cr and long-term S-Cr were age, with a positive estimation, meaning higher age is associated with a greater difference in S-Cr between baseline and long-term values. Additionally, AKI at CT was significant; these patients had a greater difference between baseline and long-term S-Cr.
Figure 8 - Baseline S-Creatinine. Density plot for baseline S-Cr (µmol/L) by hospital. Red observations received contrast at CT and blue did not.
Figure 9 - Long term S-Creatinine. Density plot for long term creatinine (µmol/L) at CT by hospital. Red observations received contrast at CT blue did not.
Figure 10 - Difference. Density plot for the difference between baseline creatinine and long-term creatinine (ln (long term/baseline)) by hospital. Red observations received contrast at CT blue did not.
Figure 8 and figure 9 show baseline and long-term S-Cr values, comparing patients exposed (red line) to contrast media with unexposed patients, between the two hospitals. Figure 10 is the most interesting one showing the logarithmic difference between baseline S-Cr and long-term S-Cr for exposed and unexposed patients, in Uppsala and in Gävle. In Gävle it seems to be a lack of decrease in observations in the exposed group, however it was not significant.
Density
0 1 2 3
20 50 100 200 500
● ●
● ●●
● ●● ● ●●●
● ●
● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●● ●● ●
Uppsala
0 1 2 3
● ●●●● ●●●●●● ● ●
● ●● ●●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ●●
● ● ●● ● ●● ● ● ●
Gävle
No ● Yes ●
Density
0.0 0.5 1.0 1.5 2.0
−1 0 1 2
● ●
●● ●
●●● ●● ●● ● ●
● ●● ● ● ●●●●●● ●●●●● ●
● ●● ● ●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●
a Uppsal
0.0 0.5 1.0 1.5 2.0
● ●
● ● ● ●
●●●●●● ● ●
● ●●●●●●●●●●●●●●●● ●
● ●● ●● ●●●●●●●
● ● ●●●●●●●●
Gävle
No ● Yes ●