Serum ketone body measurement in patients with diabetic ketoacidosis

Abstract

Background

Diabetic ketoacidosis (DKA) is a critical manifestation in patients with diabetes mellitus. DKA has been conventionally diagnosed by the presence of hyperglycemia (blood glucose levels > 250 mg/dl) and metabolic acidosis (blood gas bicarbonate [HCO₃⁻] < 18 mmol/l and pH in blood gas < 7.30). However, quantitative evaluation of serum ketone bodies has not been established. The current study investigates serum ketone body levels in patients suspected for DKA.

Methods

We have retrospectively evaluated patients with hyperglycemia whose serum ketone body levels at the outpatient clinic and emergency department were measured simultaneously with blood gas analysis during 2011–2019. Clinical backgrounds, severity of diabetes, serum ketone bodies, and blood gas factors were analyzed.

Results

Seventy-two patients were enrolled in the study, after the exclusion of patients who had ketosis due to factors other than diabetes. Serum ketone body levels were negatively correlated with the levels of blood HCO₃⁻ and pH. By receiver-operating-characteristic curve analyses, optimal cut-off values for diagnosis of DKA were determined at 6.3 mmol/l of beta-hydroxybutyrate, 1.4 mmol/l of acetoacetate, and 8.0 mmol/l of total ketone body, respectively. Moreover, serum ketone bodies appeared to effectively differentiate between type 1 and type 2 diabetes mellitus. The cut-off values were higher than those in previous reports.

Conclusions

Serum ketone body levels were thought to be useful in the diagnosis of DKA. Further investigations with increased numbers of patients are required to establish the appropriate application of serum ketone bodies in patients with DKA.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13340-022-00581-2.

Introduction

Diabetic ketoacidosis (DKA) is a severe acute complication of diabetes mellitus (DM) caused by insufficient insulin effect and characterized by the triad of hyperglycemia, metabolic acidosis and hyperketonemia [1–5].

DKA is comprehensively diagnosed by medical history, clinical manifestations, and laboratory findings. Diagnostic criteria of DKA proposed in the 2001 American Diabetes Association (ADA) guidelines were plasma glucose concentration > 250 mg/dl, blood gas bicarbonate (HCO₃⁻) level < 18 mmol/l, and blood gas pH < 7.30 [6]. Meanwhile, in the criteria of the International Society of Pediatric and Adolescent Diabetes (ISPAD), the cut-off values for DKA are plasma glucose > 200 mg/dl and blood gas HCO₃⁻ < 15 mmol/l [7]. Hyperketonemia is mentioned as a characteristic of DKA in the diagnostic criterion above [6, 7]. But the definition of ‘hyperketonemia’ is not clear; the guideline of ADA recommends the measurement of serum beta-hydroxybutyrate (βOHB) without cut-off values [6]. Moreover, the diagnostic cut-off value of serum or capillary βOHB in DKA varied in reports [7–9].

Harano, et al. previously reported serum ketone body levels in Japanese patients with type 1 DM (T1DM) and type 2 DM (T2DM) [10]. To our knowledge, there have been no investigations of ketone bodies for diagnosis of DKA in Japan. The purpose of the study was to evaluate serum ketone bodies in Japanese patients with T1DM and T2DM who had exhibited DKA.

Materials and methods

Patients

This single-center retrospective study was based in an outpatient setting (at the outpatient clinic or emergency room) at the Department of Diabetes and Endocrinology at the Japanese Red Cross Wakayama Medical Center between April 2011 and April 2019. Patients with DM (including both T1DM and T2DM) who were suspected to have DKA were enrolled in the study [6] (Supplemental Fig. 1). Briefly, (1) hyperglycemic signs include polyuria with polydipsia, weight loss, fatigue, vomiting, illness, abdominal pain, (2) symptoms of dehydration such as tachycardia, poor skin turgor, dry mucous membranes, hypotension, Kussmaul respirations, confusions, and consciousness loss, and (3) nonadherence or interruption of therapy.

Patients suspected DKA underwent measurement of serum and urine ketone bodies, then patients without simultaneous measurement of blood gas, and those who had any insulin treatment within 24 h were excluded from the study. Patients with malignancy, liver cirrhosis, severe malnutrition, hypoglycemia or alcohol abuse were excluded from the study, as were users of sodium-glucose-cotransporter-2 inhibitors. The study protocol was approved by the Japanese Red Cross Wakayama Medical Center Ethics Committee (No. 680, April 30. 2019), and all participants had the opportunity to opt-out of the research. The necessity for written informed consent from each patient was waived in this retrospective study. We used clinical information obtained in routine clinical practice, and no patients refused to participate in the study. All procedures were in accordance with the institutional ethical standards regarding human experimentation and with the Declaration of Helsinki, as revised in Fortaleza, Brazil in October 2013.

Diagnosis of DKA and DM

Diagnosis of DKA was made according to criteria from ADA: plasma glucose concentration > 250 mg/dl, blood gas bicarbonate (HCO₃⁻) level < 18 mmol/l, and blood gas pH < 7.30 [6], and T1DM was diagnosed by Japan Diabetes Society criteria [11].

Briefly, this comprised rapid development of ketosis or ketoacidosis within 3 months of the appearance of hyperglycemic symptoms, the permanent need for insulin therapy, and positive pancreatic autoantibody or absolute insulin deficiency: serum fasting C-peptide (CPR) levels < 0.6 ng/ml. Patients with T2DM were further divided into 2 groups. We have defined patients with decreased fasting CPR levels: < 0.6 ng/ml were considered to be insulin-dependent T2DM (T2DM-IDDM), and patients with CPR levels ≥ 0.6 ng/ml were considered to be non-insulin-dependent T2DM (T2DM-NIDDM).

Clinical profiles of the patients and measurement of serum ketone bodies

Clinical backgrounds, duration of DM, glycemic controls, and severity of diabetes mellitus were analyzed. Serum levels of total ketone body (TKB), acetoacetate (AcAc), and βOHB were measured by an enzymatic cycling method (Kainos Laboratories, Inc. Tokyo, Japan). The levels of serum ketone bodies were compared with the factors in blood gas analysis, such as HCO₃⁻ and pH. The measurement methods of serum ketone bodies and blood gas were unchanged between 2011 and 2019.

Statistical analysis

Clinical data among the three groups were compared by Kruskal–Wallis test (continuous variable) or chi-square test (nominal variable). To determine the ketone body values that corresponded to a HCO₃⁻ level of 18 mmol/l, serum ketone body levels from each group were compared with simultaneously taken HCO₃⁻ levels by using regression analysis. The levels of serum ketone bodies were compared with simultaneously blood levels of HCO₃⁻ or pH by Spearman’s rank correlation coefficient test. Receiver-operating-characteristic (ROC) curve analyses were used to determine cut-off values of serum or urine ketone bodies for diagnosis of DKA. P values less than 0.05 were considered to be statistically significant. Statistical analyses were performed using JMP, version15 (SAS Institute Inc., Cary, NC, USA).

Results

Among consecutive outpatients with DM, a total of 202 patients were recruited for the study. Based on the inclusion and exclusion procedures as indicated, 72 patients were enrolled in the study: 21 patients with T1DM, 29 patients with T2DM-IDDM (8 of whom could discontinue insulin therapy within a half-year), and 22 patients with T2DM-NIDDM were analyzed (Supplemental Fig. 1) (Table 1).

Table 1.

Clinical characteristics of the patients

	T1DM	T2DM			Total	1* p value
		IDDM	NIDDM	Subtotal
n (%)	21 (29.1)	29 (40.2)	22 (30.5)	51 (70.8)	72
Age (years)	48 (42–69)	61 (49–75)	50 (41–58)	54 (44–69)	54 (44–68)	0.0842
Male (%)	10 (47.6)	15 (51.7)	16 (72.7)	31 (60.8)	41 (56.9)	0.1918
BMI	19 (16–23)	22 (19–25)	28 (26–32)‡,§	25 (20–30)	23 (19–26)	< 0.0001
Duration (years)	0 (0–16)	5 (1–16)	3 (0–15)	5 (0–16)	5 (0–15)	0.3731
Glucose (mg/dl)	547 (439–824)	437 (360–624)	341 (265–536)‡	418 (302–287)	445 (341–617)	0.0032
HbA1c (%)	10.2 (8.2–11.9)	12.7 (9.4–13.5)	11.8 (9.5–14.3)	12.0 (9.6–13.5)	11.3 (9.1–13.4)	0.0505
TKB (mmol/l)	7.1 (5.3–12.6)	3.4 (0.7–8.8)†	2.9 (0.7–4.9)‡	2.9 (0.8–6.6)	4.4 (1.5–9.6)	0.0015
βOHB (mmol/l)	5.3 (3.9–9.2)	2.4 (0.5–6.4)†	2.2 (0.5–4.0)‡	2.4 (0.6–5.0)	3.3 (1.1–7.2)	0.0008
AcAc (mmol/l)	1.5 (1.0–3.1)	1.0 (0.2–2.0)†	0.6 (0.2–1.1)‡	0.6 (0.2–1.7)	1.0 (0.4–2.1)	0.0034
pH	7.27 (7.15–7.37)	7.38 (7.30–7.41)†	7.38 (7.34–7.42)†	7.38 (7.31–7.42)	7.35 (7.24–7.40)	0.0168
Bicarbonate (mmol/l)	17 (8–20)	22 (11–26)	21 (16–25)	22 (14–26)	20 (11–25)	0.1027
Lactate (mmol/l)	2.0 (1.4–3.1)	1.6 (1.1–2.8)	2.0 (1.4–3.0)	2.0 (1.4–3.1)	1.9 (1.3–2.9)	0.5782
Urine ketone (%)						0.2590
–	1 (4.8)	5 (17.2)	4 (18.2)	9 (17.6)	10 (13.9)
1 +	0 (0.0)	3 (10.3)	1 (4.6)	4 (7.8)	4 (5.6)
2 +	5 (23.8)	7 (24.1)	5 (22.7)	12 (23.5)	17 (23.6)
3 +	14 (66.7)	13 (44.8)	8 (36.4)	21 (41.1)	35 (48.6)
4 +	1 (4.8)	1 (3.5)	4 (18.2)	5 (9.8)	6 (8.3)
DKA (%)a	11 (52.4)	10 (34.5)	6 (27.3)	16 (31.4)	27 (37.5)	0.2145

Data are shown as median (lower quartile-upper quartile), or number (proportion)

1* Comparisons between the 3 groups (T1DM, T2D-IDDM, and T2D-NIDDM) were made by Kruskal–Wallis test (continuous variable) or Chi-square test (nominal variable), and p-values less than 0.05 are shown in bold

For comparison of the 2 groups, Steel–Dwass test was performed. Values shown in bold text are statistically significant. ^†p < 0.05 vs T1D; ^‡p < 0.01 vs T1D; ^§p < 0.01 vs T2D-IDDM

IDDM, insulin-dependent diabetes mellitus; NIDDM, non-insulin-dependent diabetes mellitus

^aNumber of patients with DKA and its proportion was shown. DKA has been diagnosed by the presence of hyperglycemia (blood glucose levels > 250 mg/dl) and metabolic acidosis (blood gas bicarbonate [HCO3-] < 18 mmol/l and pH in blood gas < 7.30)

Patients with T2DM-NIDDM had greater BMI compared with patients with T1DM, and compared with patients with T2DM-IDDM. Patients with T1DM had higher glucose levels than those with T2DM-NIDDM. No significant differences were seen in glycated hemoglobin (HbA1c) levels among the 3 groups. The levels of TKB, βOHB, and AcAc in patients with T1DM were significantly higher than those who were T2DM-IDDM, and also in those who were T2DM-NIDDM. In ROC analysis, 2.5 mmol/l of βOHB had 95.2% of sensitivity and 56.9% of specificity for differentiating T1DM from T2DM (Supplemental Fig. 2).

Significantly lower blood gas pH levels were observed in patients with T1DM among the 3 groups, whereas no significant differences were observed in blood gas HCO₃⁻ and lactate levels. The proportion of DKA in patients with T1DM was higher among the 3 groups, but without significance (T1DM 52.4%; T2DM-IDDM 34.5%; T2DM-NIDDM 27.5%, P = 0.2145) (Table 1).

Note that 2 patients with T2DM-NIDDM who developed DKA had no obvious episode of soft drink abuse, and showed spontaneous recovery from insulin therapy, and 12 cases of ketosis and 4 cases of ketoacidosis in T2DM-NIDDM patients were seemed to be due to excessive soft drink intake.

The levels of serum TKB, AcAc, and βOHB were all negatively correlated with the levels of blood gas HCO₃⁻ and also those of pH (Figs. 1, 2). HCO₃⁻ levels of 18 mmol/l corresponded to serum βOHB levels of 4.6 mmol/l, AcAc of 1.5 mmol/l, TKB of 6.1 mmol/l, respectively. Similarly, pH of 7.30 corresponded to βOHB of 4.6 mmol/l, AcAc of 1.5 mmol/l, TKB of 6.0 mmol/l, respectively.

Fig. 1.

Comparisons between HCO₃⁻ and blood ketone body levels. TKB (a), βOHB (b), and AcAc (c) levels were compared with blood levels of HCO₃⁻ by Spearman’s rank correlation coefficient test. Patients with T1DM (white quadrangle), T2DM-IDDM (white circle), and T2DM-NIDDM (filled circle) were analyzed separately. TKB total ketone body, βOHB beta-hydroxybutyrate, AcAc acetoacetate, HCO₃⁻ bicarbonate, T1DM type 1 diabetes mellitus, T2DM type 2 diabetes mellitus, IDDM insulin dependent diabetes mellitus, NIDDM non-insulin dependent diabetes mellitus

Fig. 2.

Comparisons between pH and blood ketone body levels. TKB (a), βOHB (b), and AcAc (c) levels were compared with blood levels of pH by Spearman’s rank correlation coefficient test. Patients with T1DM (white quadrangle), T2DM-IDDM (white circle), and T2DM-NIDDM (filled circle) were analyzed independently. TKB total ketone body, βOHB beta-hydroxybutyrate, AcAc acetoacetate, T1DM type 1 diabetes mellitus, T2DM type 2 diabetes mellitus, IDDM insulin dependent diabetes mellitus, NIDDM non-insulin dependent diabetes mellitus

In ROC analysis, 8.0 mmol/l of TKB had 81.5% of sensitivity and 100% of specificity for the diagnosis of DKA, the area under the curve (AUC) was 0.93498. Similarly, 6.3 mmol/l of βOHB had 81.5% of sensitivity and 100% of specificity, AUC was 0.93663, 1.4 mmol/l of AcAc had 88.9% of sensitivity and 88.9% of specificity, AUC was 0.92593, respectively (Fig. 3).

Fig. 3.

ROC analysis of serum ketone bodies in the diagnosis of DKA. Optimal cut-off levels of TKB (a), βOHB (b), and AcAc (c) for diagnosis of DKA were estimated by ROC analysis. a 8.0 mmol/l of TKB had 81.5% of sensitivity and 100% of specificity, area under the curve (AUC) was 0.93498. b 6.3 mmol/l of βOHB had 81.5% of sensitivity and 100% of specificity, AUC was 0.93663. c 1.4 mmol/l of AcAc had 88.9% of sensitivity and 88.9% of specificity, AUC was 0.92593. TKB total ketone body, Βohb beta-hydroxybutyrate, AcAc acetoacetate, ROC receiver-operating-characteristic

Serum βOHB levels were positively correlated with urine ketone body levels (ρ = 0.6146, p < 0.0001) (Fig. 4a) Notably, serum βOHB levels were higher in patients with DKA than in those without DKA, even at the same urinary ketone excretion level (Fig. 4b). No significant difference was observed between serum βOHB levels and urine ketone in patients with DKA (ρ = 0.2494, p = 0.2096), and 2 cases of DKA exhibited no urinary ketone excretion (Fig. 4b left). Serum βOHB levels were positively correlated with urine ketone levels in patients without DKA (ρ = 0.6417, p < 0.0001) (Fig. 4b right).

Fig. 4.

Serum βOHB levels in comparison to the levels of urine ketone body. In box-and-whisker plots, minimum, lower quartile, median, upper quartile, and maximum levels are shown as indicated. Serum βOHB levels were compared with urine ketone levels by Spearman’s rank correlation coefficient tests in all cases (a), and patients with and without DKA (b), respectively. In all cases (a), serum βOHB levels (mmol/l) in median (interquartile ranges: IQR) were 0.3 (0.1–0.4) at urine ketone (−), 0.5 (0.2–2.6) at urine ketone (+), 2.4 (1.1–4.3) at urine ketone (2 +), 5.2 (2.5–9.1) at urine ketone (3 +), 6.6 (2.3–7.9) at urine ketone (4 +). In patients with DKA (b left), serum βOHB levels (mmol/l) in median (IQL) were 2.1 (0.1–4.2) at urine ketone (−), 5.2 (3.3–13.9) at urine ketone (2 +), 9.1 (7.8–12.3) at urine ketone (3 +), 7.3 (6.4–8.3) at urine ketone (4 +). In patients without DKA (b right), serum βOHB levels (mmol/l) in median (IQL) were 0.3 (0.1–0.3) at urine ketone (−), 0.5 (0.2–2.6) at urine ketone ( +), 2.3 (0.8–3.6) at urine ketone (2 +), 2.7 (1.5–4.5) at urine ketone (3 +), 2.2 (2.0–2.4) at urine ketone (4 +). βOHB beta-hydroxybutyrate, DKA diabetic ketoacidosis

Discussion

The current study has three principal findings, (1) an association of serum ketone body levels with blood gas HCO₃⁻, and that with blood gas pH were shown in our Japanese cohort, (2) diagnostic utilities of these ketone body levels in DKA were shown with high sensitivity and specificity, (3) serum ketone bodies seemed to effectively differentiate between T1DM and T2DM. Taken together, blood ketone body levels seem to be useful for the diagnosis of DKA.

Harano, et al. investigated serum ketone body levels in Japanese patients with T1DM (n = 117), and with T2DM (n = 281), and this included 4 patients with T1DM and DKA [10]. They finally concluded that serum levels of TKB did not exceed 2.0 mmol/l in patients with T2DM, that insulin therapy would reduce serum ketone bodies in patients with T1DM, and that serum levels of ketone bodies are useful for the diagnosis of T1DM (those with TKB > 2 mmol/l) and for detecting insufficient insulin therapy. The current study further suggests the novel beneficial impact of serum ketone bodies on the diagnosis of DKA in both T1DM and T2DM.

‘Ketone body’ consists of acetone, AcAc, and βOHB, which derives from lipolysis and are utilized as fuels [2]. Ketone body production is suppressed by insulin [3]. Physiological hyperketonemia reflects the acceleration of lipolysis and ketogenesis due to insulin insufficiency, and it can also be seen in healthy individuals such as in fasting [2–4]. When AcAc and βOHB are overproduced and accumulated, metabolic acidosis is induced [2, 5]. Therefore, the ketone body is widely known as a marker of insulin action on adipose tissue, and may reflect the severity of DKA.

In comparison with previous reports on serum ketone bodies in DKA [7–9], the cut-off value of βOHB, 4.6 mmol/l in DKA in the current study is considered to be higher than that of previous studies. In the guideline of ISPAD, ketonemia is defined as blood βOHB ≥ 3mmol/l [7]. Sheikh-Ali, et al. showed a strong negative association between serum HCO₃^- and βOHB and suggested a cut-off value of βOHB ≥ 3.0mmol/l in children, and ≥ 3.8mmol/l in adults for diagnosis of DKA [8]. Arora, et al. finally concluded the cut-off value of capillary βOHB ≥ 1.5mmol/l for diagnosis of DKA [9].

The reason for this discrepancy is unclear, but possible explanations might be a difference in ethnicity, prevalence of T1DM, etiology of DKA, capacity of insulin secretion, or body composition. Importantly, higher cut-off values of ketone bodies in this study might be due to the different inclusion criteria among the studies. Moreover, the differences in the measurements of methods for both serum ketone bodies and blood gas (venous, arterial, or capillary) may affect the absolute values. Sheikh-Ali, et al. used ‘P Module of a Roche Modular Analytics System to measure βOHB’ [8], whereas Arora S, et al. used a ‘point-of-care βOHB measurement’ (Precision Xtra meter; Medisense/Abbot Laboratories) [9].

Serum ketone body levels were surely and negatively correlated with the levels of blood gas HCO₃⁻, and those of blood gas pH. ROC analyses revealed that optimal cut-off values for diagnosis of DKA were determined at 6.3 mmol/l of βOHB, 1.4 mmol/l of AcAc, and 8.0 mmol/l of total ketone body, respectively. Therefore, from a practical point of view, these cut-off levels may be used in the diagnosis of DKA.

Significantly higher levels of TKB, βOHB, and AcAc, and lower blood gas pH were observed in patients with T1DM than in patients with T2DM-IDDM and T2DM-NIDDM. Therefore, serum ketone bodies appeared to effectively differentiate between T1DM and T2DM. No significant differences were observed in blood gas HCO₃⁻ levels among the three groups, although the reasons for this remain unknown.

Diagnostic power in DKA based on the conventional criteria for DKA and the current study which includes ketone body-based diagnosis appeared to be similar [7]. In addition to the diagnosis of DKA, the current study might have advantages in the diagnosis and therapeutic strategy for certain subtypes of DM other than typical T1DM or T1DM. First, regarding atypical T2DM along with ketosis or ketoacidosis onset, known as ketosis-prone T2DM (KPD) [12, 13]. These patients are characterized by higher BMI, being negative for islet-cell autoimmunity, spontaneous recovery from β-cell dysfunction, and a predominance among African–American individuals [12]. Aizawa et al. reported 5 Japanese patients with KPD, and the duration of insulin dependency was shorter than 4 weeks [13]. Two T2DM-NIDDM patients with DKA in the current study had no obvious episode of soft drink abuse, required insulin therapy, and subsided. Therefore, they might be diagnosed with KPD. However, future studies are required to diagnose them definitely as KPD by definition of appropriate cut-off levels of serum ketone bodies.

There is another type of T2DM so-called ‘soft-drink ketosis’, where acute deterioration of hyperglycemia is driven by excessive intake of soft drinks containing sugar, which can give rise to ketosis or ketoacidosis [14]. Patients with T2DM-NIDDM and high BMI are often diagnosed as T2DM due to ‘soft-drink ketosis’. Our data had 12 cases of ketosis and 4 cases of DKA due to excessive soft drink intake. The serum ketone body levels in the cases of ‘soft-drink ketosis’ remain unknown, and further investigations are needed to clarify the role of serum ketone bodies in ‘soft-drink ketosis’.

Our data indicated a practical limit of urine ketone in the diagnosis of DKA, which is in concordance with previous reports [2, 9, 15]. In addition to the low specificity of urinary ketone measurements, serum ketone bodies are more reliable than urine ketone bodies for the detection of DKA.

Our study has some limitations. First, sampling bias is inevitable because of the retrospective study design. Second, counter-regulatory hormones were not measured. Third, the results of serum ketone bodies cannot be obtained immediately in most of the hospitals, unfortunately. Fourth, meal statuses were unclear, although most patients had a loss of appetite. Fifth, SGLT2i user was excluded from this study because the number of those was limited. Taken together, DKA should be managed comprehensively based on complex examinations including usual laboratory examination, serum ketone bodies, and blood gas tests. In addition to the diagnosis of DKA, an indication of intensive insulin therapy needs optimal cut-off value of serum ketone body levels. Future studies of serum ketone body levels in relation to therapy (with or without SGLT2i) and prognosis for DM should be considered.

In conclusion, serum ketone body might be useful in the diagnosis of DKA, as well as serum HCO₃⁻ and pH. Further investigations with a greater number of patients are required to establish the cut-off levels of serum ketone body levels in the management of DKA.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

AcAc

Acetoacetate

CPR

C-peptide

DM

Diabetes mellitus

DKA

Diabetic ketoacidosis

GADAb

Glutamic acid decarboxylase autoantibody

HCO₃⁻

Bicarbonate

ICAs

Autoantibodies to islet cells

IDDM

Insulin-dependent diabetes mellitus

KPD

Ketosis-prone type 2 diabetes mellitus

NIDDM

Non-insulin-dependent diabetes mellitus

ROC

Receiver-operating-characteristic

TKB

Total ketone body

T1DM

Type 1 diabetes mellitus

T2DM

Type 2 diabetes mellitus

βOHB

Beta-hydroxybutyrate

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data availability statement

The datasets in the manuscript are available from the corresponding authors on reasonable request.

Declarations

Conflict of interest

The authors have no competing interests to disclose.

Compliance with ethical standards

The study protocol was approved by the Japanese Red Cross Wakayama Medical Center Ethics Committee (No. 680, April 30. 2019), and all participants had the opportunity to opt-out of research. The necessity for written informed consent from each patient was waived in this retrospective study. We used clinical information obtained in routine clinical practice, and no patients refused to participate in the study. All procedures were in accordance with the institutional ethical standards regarding human experimentation and with the Declaration of Helsinki, as revised in Fortaleza, Brazil in October 2013.