SEARCH

SEARCH BY CITATION

Keywords:

  • Anion gap;
  • Hypoproteinemia;
  • Strong ion difference;
  • Strong ion gap

Abstract

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. References

Background

The quantitative effect of strong electrolytes, pCO2, and plasma protein concentration in determining plasma pH and bicarbonate concentrations can be demonstrated with the physicochemical approach. Plasma anion gap (AG) and strong ion gap (SIG) are used to assess the presence or absence of unmeasured anions.

Hypotheses

The physicochemical approach is useful for detection and explanation of acid-base disorders in horses with colitis. AG and SIG accurately predict hyperlactatemia in horses with colitis.

Animals

Fifty-four horses with acute colitis and diarrhea.

Methods

Retrospective study. Physicochemical variables were calculated for each patient. ROC curves were generated to analyze sensitivity and specificity of AG and SIG for predicting hyperlactatemia.

Results

Physicochemical interpretation of acid-base events indicated that strong ion metabolic acidosis was present in 39 (72%) horses. Mixed strong ion acidosis and decreased weak acid (hypoproteinemia) alkalosis was concomitantly present in 17 (30%) patients. The sensitivity and specificity of AG and SIG to predict hyperlactatemia (L-lactate > 5 mEq/L) were 100% (95% CI, 66.4–100; P < .0001) and 84.4% (95% CI, 70.5–93.5 P < .0001). Area under the ROC curve for AG and SIG for predicting hyperlactatemia was 0.95 (95% CI, 0.86–0.99) and 0.93 (95% CI, 0.83–0.99), respectively.

Conclusion and Clinical relevance

These results emphasize the importance of strong ions and proteins in the maintenance of the acid-base equilibria. AG and SIG were considered good predictors of clinically relevant hyperlactatemia.

Abbreviations
A

total net negative charge of blood proteins

AG

anion gap

Atot

total plasma concentration of nonvolatile weak acids

BE(ecf)

base excess extracellular fluid

H-H

Henderson-Hasselbalch approach

HCO3

bicarbonate

mEq

milliequivalent

pCO2

partial carbon dioxide pressure

SID

strong ion difference

SIG

strong ion gap

TP

total protein

Historically, 2 approaches have been used to evaluate acid-base imbalance, the Henderson-Hasselbach approach (H-H)[1] and more recently the physicochemical approach.[2, 3] The H-H currently focuses on plasma bicarbonate [HCO3] and extracellular base excess (BE(ecf)) concentrations as indicators of the metabolic components of acid-base balance. The H-H approach is clinically adequate to describe acid-base disorders. However, it is more descriptive than mechanistic and ignores the importance of electrolyte and protein concentrations in the determination of acid-base balance.[4, 5] The quantitative physicochemical approach emphasizes the importance of strong electrolytes (Na+, K+, Cl, L-lactate), pCO2, and total plasma protein concentration (Atot) in determining plasma pH and [HCO3].[2] This physicochemical approach is mechanistic and provides a more in-depth insight into the pathophysiology of acid-base imbalance.

Horses with acute diarrhea develop intraluminal sequestration of fluid, profuse watery diarrhea, severe electrolyte abnormalities, and often also protein loss.[6] In addition, hypoperfusion because of dehydration may lead to tissue hypoxia and anaerobic metabolism. Traditionally, L-lactate has been viewed as a marker of hypoperfusion and is considered an accurate predictor of mortality in different species.[7] Clinicians assume that another commonly calculated acid-base parameter, the anion gap (AG), is a sufficient alternative to use in predicting hyperlactatemia. However, several studies have shown conflicting results regarding the sensitivity and specificity of AG when investigating metabolic acidosis in humans, horses, and cattle.[7-12] On the other hand, SIG has shown an excellent correlation with L-lactate concentration in healthy horses under condition of exercise,[13] as well as in critically ill horses.[10, 14] However, there is a lack of information regarding the sensitivity and specificity of AG and SIG to predict [L-lactate] in critically ill adult horses with abnormal concentrations of electrolytes and plasma proteins because of gastrointestinal disease. The objectives of this study were to investigate the acid-base status in horses with acute diarrhea by the physicochemical approach and to assess whether AG and SIG can be used as screening tools for hyperlactatemia in horses with colitis.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. References

Study Population

Retrospectively, medical records of horses admitted to the Ontario Veterinary College University of Guelph between January 2008 and December 2010 that were presented for acute diarrhea and had a diagnosis of typhlitis, typhocolitis, or colitis were reviewed. All horses included in the study met all of the following inclusion criteria: >1 year of age, duration of diarrhea (increased frequency and water content in the feces) < 24 hours, measurement of blood gases and basic serum metabolic profile within the 1st hour of hospitalization before to parenteral administration of crystalloids.

Data Collection

For each patient, age, breed, sex, treatment with antibiotics or nonsteroidal anti-inflammatory drugs (NSAIDs) before the onset of diarrhea, and long transport or previous surgery requiring general anesthesia were recorded. From the physical examination, mentation status, heart rate (HR), respiratory rate (RR), presence of a toxic line in the oral mucous membranes, rectal temperature (T), and signs of phlebitis, thrombophlebitis, laminitis or peritonitis were recorded. Clinical endotoxemia was considered present in a horse if ≥ 3 of the following abnormalities were identified: tachycardia, fever or hypothermia, presence of a toxic line in the oral mucous membranes, leukopenia or leukocytosis. Medical complications during hospitalization were recorded. Survival and duration of hospitalization were recorded on the day that the horse was discharged from the hospital or on the day of euthanasia or death. Survival was defined as discharge from the hospital.

Measurement Techniques

A venous blood sample was collected anaerobically into a polypropylene syringe containing lyophilized lithium-heparin for blood gas,1 electrolytes, and pH analyses. with a Radiometer 800 Flex blood gas machine.2 Total protein concentration was measured as total solids by refractometry.3 Clinicopathologic data of pvCO2 (mmHg), pH, and plasma [Na+] (mEq/L), [K+] (mEq/L), [Cl] (mEq/L), [L-lactate] (mEq/L), and [TP] (g/L) were recorded.

Calculations

The Henderson-Hasselbalch (H-H) equation was used to calculate [HCO3] (mEq/L) and BE(ecf) (mEq/L) from the measured values for venous pH and pCO2 and assigned values for S of 0.0307 mmol/(L × mmHg) and pK1′ of 6.095.[4] Anion gap (mEq/L) also was calculated.

Bicarbonate [HCO3][4]

  • display math(1)

[Base excess of extracellular fluid (BE(ecf))[4]

  • display math(2)

BE(ecf) assumes a fixed hemoglobin concentration of 50 g/L.[4]

Anion gap (AG)[15]

  • display math(3)

Physicochemical variables SID, A- and SIG (mEq/L), were calculated as follows:

Strong ion difference was calculated by 4 strong ions (SID4)[2]

  • display math(4)

Total negative charge of plasma concentration of weak acids (A)[3]

  • display math(5)

where Atot is the total plasma concentration of nonvolatile weak acids and pKa (6.65) is the effective dissociation constant of plasma weak acids.

SIG was calculated by the simplified strong ion model that has been applied experimentally to horse plasma and provides species-specific values of pKa and Atot.[4, 13]

  • display math(6)

Acid-Base Disorder Definitions

By the H-H equation and physicochemical approach, the acid-base disorders were defined as follows: metabolic acidosis when BE(ecf) and [HCO3] were below the reference range (BE(ecf) < 1 mEq/L and [HCO3] < 24 mEq/L); strong ion acidosis when SID4 was below the reference range and weak acid acidosis when A was above the reference range (SID4 < 38 mEq/L and A >16 mEq/L); metabolic alkalosis when BE(ecf) and [HCO3] were above the reference range (BE(ecf) > 7 mEq/L and [HCO3] > 28 mEq/L); strong ion alkalosis when SID4 was above the reference range and weak acid alkalosis when A was below the reference range (SID4 > 44 mEq/L and A < 12 mEq/L); respiratory acidosis and respiratory alkalosis when the pvCO2 was above or below the reference range (pvCO2 38–49 mmHg), respectively; hyperlactatemia and severe hyperlactatemia were defined as plasma [L-lactate] > 2 mEq/L and > 5 mEq/L, respectively.[7]

Statistical Analysis

Data were expressed as mean ± SD. Non-normally distributed data were expressed as median and lower and upper range. Normality of the data was tested by the Kolmogorov–Smirnov test. A simple linear regression was used to evaluate the relationship between plasma [L-lactate] and AG and SIG. ROC curve analysis was used to identify the AG and SIG cut-off that optimized diagnostic sensitivity and specificity for predicting severe hyperlactatemia (L-lactate > 5 mEq/L). Optimal cut-off values were defined by the points representing the highest concomitant sensitivity and specificity. Differences in the areas under the ROC curve (AUC) of AG and SIG were compared by a non-parametric method.[16] Differences were classified as significant if P-value < .05. Statistical analysis was performed by a statistical software.4

Results

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. References

Study Population

Fifty-four of 119 horses that presented for acute diarrhea and had a diagnosis of typhlocolitis during the study period met the inclusion criteria. Breed distribution was as follows: 16 (30%) Thoroughbreds, 10 (18%) Standardbred, 6 (11%) Paint Horse, 6 Warmbloods (11%), 4 (7%) Morgan, 3 (6%) Clydesdale, 3 (6%) Quarter Horses, 2 (4%) Oldenburg, and 4 (7%) horses of other breeds. The study included 26 (48%) females, 12 (22%) intact males and 16 (30%) geldings. The median age of the horses was 5.5 (range, 1–33). Fifteen (28%) and 13 (24%) of the animals had a previous history of antimicrobial or NSAIDs treatment, respectively. Four (7%) had received antimicrobials and NSAIDs simultaneously. Two (4%) horses had a history of long-duration transportation and 2 (4%) animals had surgery under general anesthesia within 1 week before development of diarrhea.

Physical Examination Findings and Outcome

At the time of hospitalization, 28 (52%) horses had tachycardia (HR> 60 bpm), 13 (24%) had tachypnea (RR > 24 rpm), 11 (20%) had fever (T°> 38.5°C), and 4 (7.4%) had hypothermia (T°< 37°C). Dehydration was estimated to be 5–8%, 8–10%, and 10–12% in 10 (19%), 35 (65%), and 9 (17%) horses, respectively. Clinical endotoxemia was considered in 18 (33%) patients. On admission, 5 (9%) horses had signs of phlebitis or thrombophlebitis, 4 (7%) had signs of laminitis, and 1 (2%) had acute renal failure. Hospital complications included 5 (8%) cases of thrombophlebitis and 1 (2%) case of laminitis. Forty-one (76%) horses were discharged, whereas 13 (24%) horses were euthanized. Four of 5 horses that developed laminitis were euthanized because of grave prognosis or financial constraints.

Acid-Base and Electrolyte Abnormalities

Thirty-six (67%) horses were hyponatremic ([Na+] < 132 mEq/L), 41 (76%) were hypokalemic ([K+] < 2.4mEq/L), 12 (22%) were hyperchloremic ([Cl] > 109mEq/L), and 28 (52%) were hypochloremic ([Cl] < 99 mEq/L). In addition, 24 (44%) were hyperlactatemic ([L-lactate] > 2 mEq/L), 36 (67%) were hypoproteinemic ([TP] < 54 g/L), 11 (20%) had increased AG (AG > 16 mEq/L), and 20 (37%) had increased SIG (SIG < −2 mEq/L). The values of the venous blood gas analysis, plasma concentration of electrolytes and H-H, and physicochemical variables of the 54 horses are presented in the Table 1. Application of the H-H approach for diagnosis of acid-base disturbances indicated that there were 22 (41%) and 27 (50%) horses with metabolic acidosis based on BE(ecf) (BE(ecf) < 1 mEq/L) and plasma [HCO3] ([HCO3] < 24 mEq/L), respectively. Twenty-one (77%) of 27 horses were diagnosed with metabolic acidosis by BE(ecf) also had decreased concentrations of plasma [HCO3]. The H-H approach also indicated that 11 (20%) of the horses had anion gap acidosis (AG > 16 mEq/L). The interpretation of the acid-base disorders on admission of each horse by the H-H approach is presented in Figure 1. Physicochemical interpretation of acid-base events indicated that strong ion acidosis (SID4 < 38 mEq/L) was present in 39 (72%) horses, and that none of the horses had strong ion alkalosis. The decreases in SID4 were primarily because of hyponatremia and increased concentration of the strong ion L-lactate. Nine of 39 (24%) horses with SID4 acidosis had hyponatremia, 5 (13%) were hyperlactatemic and 12 (31%) had both hyponatremia and hyperlactatemia. Decreased weak acid (hypoproteinemia) alkalosis (A < 12 mEq/L) was diagnosed in 19 patients. Mixed strong ion acidosis and decreased weak acid alkalosis was present concomitantly in 17 (30%) of the patients. Mild respiratory acidosis and respiratory alkalosis were detected in 2 (4%) and 10 (18%) horses, respectively. The interpretation of acid-base disorders of each horse on admission by the physicochemical approach is presented in Figure 2.

Table 1. Acid-base parameters of 54 horses with acute colitis and diarrhea obtained on admission
ParameterHorsesRef. Value[10, 13, 39]
  1. Data presented as mean ± SD.

  2. a

    Non-normal distributed data presented as median, upper, and lower ranges.

pH7.38 ± 0.057.32–7.44
PCO2 (mmHg)40.7 ± 4.938–49
HCO3 (mEq/L)23.9 ± 3.924–28
BE(ecf) (mEq/L)-0.4 ± 4.41.1–7.1
Cl (mEq/L)94 ± 6.799–109
SID4 (mEq/L)35.3 ± 4.438–43
TP (g/L)56 ± 1454–79
A (mEq/L)12.6 ± 3.312–16
aNa+ (mEq/L)130 (124–145)132–146
aK+ (mEq/L)3.2 (2.8–3.4)2.4–4.7
aLac (mEq/L)1.8 (1.4–13.5)0–2
aAG (mEq/L)14 (10.5–16.7)6–16
aSIG (mEq/L)−0.7 (−3.6–1.6)−2–2
image

Figure 1. Interpretation of acid-base disorders on admission for each horse by H-H approach. Metabolic acidosis was determined by means of plasma concentration of bicarbonate ([HCO3]) (A) and base excess extracellular fluid [BE(ecf)]) (B).AG anion gap; pCO2 partial carbon dioxide pressure.

Download figure to PowerPoint

image

Figure 2. Interpretation of acid-base disorders on admission for each horse by the physicochemical approach. SIG strong ion gap; pCO2 partial carbon dioxide pressure; SID strong ion difference; A total net negative charge of blood proteins.

Download figure to PowerPoint

Sensitivity and Specificity of AG and SIG for Predicting Hyperlactatemia

Calculated AG and SIG were significantly associated with plasma [L-lactate], AG = 10.4 + 1.36 ×  [L-lactate] (R2 = 0.44; P < .0001), SIG = 1.90−1.17 ×  [L-lactate] (R2 = 0.48; P < .0001). The areas under ROC analysis of AG and SIG for prediction of severe hyperlactatemia were 0.95 (95% CI, 0.86–0.99) and 0.93 (95% CI, 0.83–0.98), respectively. Comparison between area under ROC analysis of AG and SIG for predicting hyperlactatemia (AUC 0.95 versus 0.93; P = .46) did not show significant difference (Fig 3). The ROC analysis identified that a cut-off of AG > 16 mEq/L and SIG < −2 mEq/L had a sensitivity of 100% (95% CI, 66.4–100; P < .0001) and specificity of 84% (95% CI, 70.5–93.5 P < .0001) for predicting severe hyperlactatemia.

image

Figure 3. Receiver operator curve (ROC) analysis of AG and SIG for predicting hyperlactatemia. Area under the ROC for AG (solid) 0.956 (95% CI: 0.862–0.993) and SIG (dashed): 0.938 (95% CI: 0.838–0.986).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. References

Strong ion acidosis and weak acid (hypoproteinemic) alkalosis were the most common acid-base imbalances in the group of horses included in this study and were simultaneously present in 30% of the patients. Hyponatremia, normochloremia or hyperchloremia, hyperlactatemia, and hypoproteinemia were considered the abnormalities responsible for these acid-base disorders. Stewart's major contribution to clinical acid-base physiology was his proposal that plasma [H+] is determined by 3 independent factors: pCO2, [SID], and Atot. An understanding of these 3 variables is required to apply the strong ion approach to acid-base disorders in plasma.[3]

This study showed that strong ion acidosis (decreased SID4) was the most common acid-base disorder in horses with acute colitis and diarrhea. Theoretically, SID may be affected in several ways to produce metabolic acidosis such as a decrease in [Na+], an increase in [Cl], endogenous production of strong anions (lactic acidosis), and impaired excretion of unmeasured anions.[17, 18] Plasma electrolyte disorders including hyponatremia and hyperlacatemia, in the presence of normochloremia or hyperchloremia, were observed frequently and likely were the cause of the decreased plasma SID4. The important role that hyponatremia plays in producing acidemia and strong ion acidosis (decreased SID) was reported in calves with diarrhea,[19] and it was attributed to excessive loss of sodium into the gastrointestinal tract and decreased milk intake because of anorexia.[20] In an experimental model of castor oil-induced diarrhea in horses, 64–74% of the sodium loss was because of the sodium content of the feces.[21] This marked increase in sodium loss suggests 2 possibilities, firstly, the sodium present in the cecal and colonic fluid was not absorbed because of damage to the absorptive surface, with further losses of sodium in the feces as a result of hyperperistalsis.[22] Secondly, some pathogens' toxins, such as salmonella outer protein B (SopB) of Salmonella spp.[23], B toxin of C. difficile and α toxin C. perfringens, may activate cellular mechanisms that elicit fluid and electrolyte secretion into the intestinal lumen.[24] The presence of hyponatremia with normochloremia or hyperchloremia may be another factor contributing to decrease in SID4 in this group of horses. Indeed, in an experimental model of diarrhea, fecal fluid [Cl] did not increase, whereas urinary excretion of [Cl] was decreased, resulting in a preservation or increase in plasma [Cl] that contributes to the development of strong ion acidosis.[21]

As stated above, SID also may be decreased by an increase in strong anions such as L-lactate. The existence of clinical sings of endotoxemia in a high proportion of horses included in this study, accompanied by moderate to severe dehydration, may contribute to and explain the presence of increased L-lactate concentrations resulting in strong ion acidosis. Excessive losses of sodium-containing fluids associated with acute diarrhea result in a decrease in effective circulating volume, which, if severe enough, may result in inadequate tissue perfusion and development of systemic lactic acidosis.[25] One of the most important consequences of endotoxemia is the compromise of organ perfusion because of arterial hypotension and inadequate tissue perfusion. Furthermore, endotoxin impairs oxygen extraction by tissues.[26] As a result, anaerobic glycolysis, combined with direct inhibition of pyruvate dehydrogenase by endotoxin generates severe local lactic acidosis.[27, 28] Interestingly, circulating endotoxin has been detected in approximately 25–40% of horses with colic.[29, 30] We hypothesized that the reason for the SID4 acidosis in the horses included in this study included increased loss of fluid with high sodium and low chloride concentrations in the diarrhea, decreased renal chloride excretion, and increased plasma L-lactate concentrations caused by dehydration and endotoxemia. These findings emphasize the importance of strong ions in the maintenance of the acid-base balance and indicate that replacement fluid therapy in horses with diarrhea should be aimed at resolving volume deficits, improving peripheral perfusion and correcting strong ion imbalances.[31]

Weak acid alkalosis also was a common acid-base disorder in these horses with acute colitis. The role of plasma protein concentration in acid-base balance is well recognized in human and veterinary medicine, with hypoalbuminemia and hyperalbuminemia causing alkalemia and acidemia, respectively.[32] The etiology of hypoalbuminemia in critical illness is complex and may involve a number of mechanisms such as an imbalance between albumin synthesis and degradation, increased capillary leakage, and altered intravascular and tissue albumin distribution.[33] In horses with colitis, the inflammation and ulceration of the mucosa lead to protein loss, predominantly low-molecular-weight proteins such as albumin.[6] Despite the fact that critically ill human and equine patients often present with hypoalbuminemia and that it is recognized that the electric charge of albumin modifies the [H+] in the plasma, there is no evidence that clinicians should treat hypoalbuminemia as an acid–base disorder. In fact, currently, there is no evidence that the body regulates Atot to maintain acid-base balance.[34]

The ROC curves analysis demonstrated that a cut-off of AG > 16 mEq/L and SIG < −2 mEq/L predict the presence or absence of clinically relevant hyperlactatemia. The AG was designed specifically to detect unmeasured anions.[15] Several studies have showed conflicting results about the sensitivity and specificity of AG when investigating metabolic acidosis.[8, 10, 11, 13, 35] Rocktaeschel et al showed that AG is reasonably accurate in predicting lactate concentrations > 5 mmol/L in critically ill humans.[8] Constable et al found that AG is accurate and clinically useful to estimate unmeasured anions in horses with normal protein concentrations.[13] In critically ill foals, a moderate correlation has been shown to exist between AG and plasma [L-lactate], with AG having a good sensitivity (77%) and specificity (83%) in detection of hyperlactatemia >5 mmol/L,[10] whereas Chawla et al found that AG is not valuable to diagnose the presence of mild hyperlactatemia because of the susceptibility of AG to changes in albumin concentrations.[35] Moreover, AG failed to provide a specific indicator of blood [L-lactate] concentration in critically ill cattle.[11] The area under ROC analysis showed that AG was a reliable screening tool for clinically relevant hyperlactatemia in this group of horses with acute colitis and diarrhea, regardless of the presence of hypoproteinemia. The reference range of AG traditionally has been defined as 6–16 mEq/L. This wide reference range occurs because 4 different measured ions ([Na+], [K+], [Cl] and [HCO3]), with their own inherent reference range, are used for the calculation. However, because we used a high plasma [L-lactate] cut-off (5 mEq/L) to define severe hyperlactatemia, we ensured that the relative concentration of this unmeasured anion would not permit the AG to remain within the reference range during hyperlactatemia.[36]

SIG represents the concentration of unmeasured strong ions in plasma and is a logical extension of the AG.[4] Constable et al found a strong correlation between the simplified SIG and plasma [L-lactate] in horses,[13] whereas in critically ill foals, SIG correlated moderately with [L-lactate] and was highly correlated with AG.[10] These results are in accordance with the results of this study. The area under ROC analysis indicated that when the threshold of plasma [L-lactate] was 5 mEq/L, the sensitivity and the specificity of SIG were excellent. In human patients with lactic acidosis, as well as patients with unexplained acidosis with normal or near normal blood [L-lactate], the plasma concentrations of acids usually associated with the Krebs tricarboxylic acid cycle are significantly increased.[37] The accumulation of such acids may contribute to the increases of the AG and SIG and account, in part, for the unmeasured anions in patients with strong ion acidosis. A study evaluating the concentration of L-lactate, pyruvate, D-lactate, acetoacetate, and 3-hydroxybutyrate concentrations in horses with intestinal disorders showed that pyruvate was increased in those horses, but that the concentrations were < 10% of L-lactate concentrations. The concentration of D-lactate was increased in half of the horses, with the highest concentration measuring 2.3 mEq/L.[38] These results suggest that unmeasured strong anions other than L-lactate are unlikely to be present in high concentrations in horses with gastrointestinal disorders. However, this speculation remains to be proven in a prospective study of horses with different diseases causing acidosis.

This study had several limitations, most notably its retrospective design, as well as the highly selected population of horses with acute colitis, which would tend to bias the study toward sicker patients. However, such horses seem to be representative of the population presented to referral teaching hospitals. Furthermore, the number of subjects was small, and to maximize the validity of our results, a larger population would be required.

Acknowledgment

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. References

Conflict of Interest: Authors disclose no conflict of interest.

Footnotes
  1. 1

    Marquest-Gaslyte vital signs, Inc; Englewood, CO

  2. 2

    Radiometer ABL 800 Flex, Radiometer medical APs, Bronshoj, Denmark

  3. 3

    Master SUR/Nα. Atago U.S.A. Inc, Bellevue, WA

  4. 4

    MedCalc, Software, Mariakerke, Belgium

References

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. References