Usefulness of a Diagnostic Algorithm to Diagnose Thrombotic Microangiopathies in Pregnancy

Introduction Maternal mortality is an important indicator of a country level of development. Maternal death is considered as a death in pregnancy at any gestational age or up to 42 days postpartum, either by direct or indirect causes. The United Nations in its Millennium Development Goals proposed the reduction of the maternal mortality ratio by three quarters by 2015 in every country worldwide. Colombia has made great progress to reducing mortality from direct causes, but not in deaths from indirect causes. In part, the problem is the large knowledge gap in the incidence and characteristics of these diseases in middle-income countries such as Colombia, where very little is known and research is scant.

aHUS is defined as an ultra-rare disease, thus, it is not usually considered as a differential diagnosis. This disease often progresses to end stage renal disease despite standard treatment with plasma exchange, with a high mortality. In recent years, the role played by the complement system in the induction of endothelial damage in patients with aHUS has been established by characterizing multiple mutations and polymorphisms in genes encoding certain complement factors. Recently, treatment with Eculizumab (monoclonal antibody that inhibits the terminal moiety complement blocking complex formation membrane attack) in prospective studies in patients with aHUS has demonstrated rapid and sustained interruption of TMA, with a significant improvement of long-term renal function and a significant reduction in requirement of renal replacement therapy (RRT) or plasmapheresis.

The literature reports that the prevalence of aHUS is higher in adult women than in men, and that 21% of all cases in women occur related to pregnancy. However, the approach and treatment of this group represents a diagnostic and treatment challenge, because many of the clinical pictures are similar to other well-defined disorders in the obstetric population, with a high risk of misdiagnosis, especially with HELLP syndrome (hemolysis, elevated liver enzymes and low platelet).

Justification and Problem Statement The challenge of the clinician in obstetric patients is to identify and assess a differential diagnosis, with both non-specific symptomatology and laboratory results, resulting in a narrow line dividing the different TMAs, in this case, aHUS with other pathologies related to pregnancy such as HELLP syndrome, Acute Fatty Liver of Pregnancy (AFLP), sepsis and preeclampsia, as well as other microangiopathies that share many of clinical characteristics of aHUS and can consequently generate a diagnostic challenge for pregnant and postpartum women.

For this, a model or algorithm leading to an accurate diagnosis that minimizes the chances of incorrect identification of pathologies associated with TMA that can be triggered during pregnancy and postpartum is needed. This diagnostic algorithm for TMA in pregnant women will have an impact on improvements on health and could have a global health benefit. Currently, two consensus documents (from Spain and Colombia), present a systematic approach and a useful diagnostic flowchart for patients suspected with a TMA. The Colombian consensus flowchart was developed in December 2014 and published in August 2015. However, these diagnostic approaches have never been validated in the obstetric population. Therefore, the following research question is presented:

Does the proposed diagnosis flowchart of thrombotic microangiopathies by the Colombian consensus result in proper classification in critically ill obstetric patients?

Objectives General objective

• To estimate the incidence of TMAs in obstetric, critically ill, patients using the consensus flowchart proposed in Colombia.

Specific objectives

• To compare the TMAs incidence in obstetric critically ill patients according to the Colombian consensus flowchart compared with expert judgment/opinion, in a single-center critical care cohort of pregnant and postpartum women from Cartagena, Colombia.
• To estimate the degree of agreement (i.e. Kappa index) of the two classifications (by the Colombian consensus flowchart and expert judgment) in obstetric critically ill patients.
• To estimate the risk of complications and adverse outcomes (death, mechanical ventilation, vasopressor support, hysterectomy, transfusions, prolonged length of stay in critical care, and multiple organ dysfunction) in obstetric critically ill patients with aHUS, (Colombian consensus flowchart and expert judgment).
• Estimate survival at 30 days after discharge in obstetric critically ill patients with aHUS, according to two classifications studied (the Colombian consensus flowchart and expert judgment) and determine if there are variables that can indicate degree of severity of the disease.

Proposed methodology A retrospective cohort study of critically ill obstetric patients. For this purpose, two classifications of TMAs will be considered and assessed in pregnant and postpartum patients, one based in the Colombian consensus algorithm of TMAs, and the other diagnostic approach based on clinical findings of highly qualified experts selected uniquely for this purpose that will be considered as the gold standard.

Eligible study population All obstetric admissions registered between 2006 and 2011 in a medical-surgical ICU in a tertiary center (Gestion Salud clinic), in Cartagena, Colombia, with about 8000 deliveries per year, will be included in this study.

Sample size estimation to detect agreement/disagreement Based on a very low prevalence of TMAs in general population, but with an increased frequency in pregnant women (1 per 1000 deliveries, estimated from the Oklahoma TTP Registry data), looking for a low level of agreement (null value) better than moderate (0.6) with a power of 80% and a clinically acceptable level of agreement (effect size) of 0.9 and a proportion of positive ratings of 0.7, the sample size should be of 66 cases for TMAs, but expecting some missing data, the sample should be at least of 20% more, thus the sample size should be of 75 cases.

Database identification and patient's selection criteria

Patients will be included if admitted to the ICU with a diagnosis of hypertensive disorder associated with pregnancy and/or sepsis, and meet the following criteria for pregnancy related thrombotic microangiopathy: thrombocytopenia <150.000 cells/mL plus non-immune hemolytic anemia (Hb less than 11gm/L plus Lactate dehydrogenase (LDH) > 600 U/L), thus the following algorithms will be established for the two main diseases proposed by the Colombian consensus algorithm:

• Thrombotic Thrombocytopenic Purpura (TTP): platelets <30.000 cells/mL plus neurological impairment, or without neurological impairment, but in addition to a creatinine level of <1.7mg/dl.
• Atypical Hemolytic Uremic Syndrome (aHUS): platelets count >30.000 cells/mL, plus creatinine > 1.7mg/dl with or without neurological impairment, or another organ dysfunction per WHO dysfunction´s criteria.
• Other diagnoses will also be considered using standard definitions, such as Acute Fatty Liver of Pregnancy (AFLP) and HELLP syndrome. Finally, the patients not classified in any of the above, shall be defined as severe preeclampsia or sepsis, as appropriate and according to standard international definitions.

Two (2) groups of experts will be defined. The first group (Group 1) shall consist of two (2) Nephrologist and two obstetricians (JC, JN, JT, RB), with extensive experience in the approach of MAT and aHUS. Each expert will separately assess the data. Meanwhile the second group (defined as Group 2), will consist of two (2) intensivists (JARS, GO), this group will apply the Colombian consensus flowchart without any other clinical guidance, on the patients from the ICU database using a pathway already described by consensus on the retrospective sample.

Data analysis All data will be collected in a database of Microsoft Excel©, and analyzed in Stata 14 statistical package (Stata 14 for Windows; StataCorp, College Station, TX, USA). A P<0.05 will be considered significant in all analyzes.

Descriptive data of the study groups, stratified by TMA diagnosis, will be provided in summary tables. Continuous data will be reported as mean ± standard deviation (SD) or median with interquartile range (IQR), depending on data normality. The normality of the variables will be evaluated by the Shapiro-Wilk test. Categorical data will be reported in percentages.

Diagnostic agreement between the two groups (Colombian consensus flow chart versus expert opinion) will be evaluated. The cumulative incidence of each of the classifications will be reported and compared using the Kappa index. The Kappa index is a measure of agreement beyond chance. A value <0 indicates no agreement and a value between 0.41 and 0.60 is indicative of a moderate agreement. Values above 0.60 are considered very good or excellent.

The rates of adverse outcomes between TMA groups (TTP or aHUS) will be compared using chi-square test for categorical data or t-test for continuous data. Non-normal data will be compared through non-parametric test of medians. A crude and adjusted multivariable logistic regression will be carried out to determine independent risk factors for adverse outcomes by TMA diagnosis. Data will be presented as unadjusted or adjusted odds ratios with 95% confidence intervals. Kaplan-Mayer will be also calculated for the two classifications of aHUS in the study population.

Afterwards, an agreement analysis of each patient will be assessed between group 1 and 2. Patients classified differently by the two groups (defined as overlap, overlapping or imitator diagnoses), will be analyzed independently to find the peculiarities that led to that situation. In case of no agreement, the conventional clinical approximation and diagnosis performed by group 1 will be considered as the gold standard.

In the final analysis, the primary and the secondary outcomes, as well as any proposals will be exposed in a meeting including all the authors and adjustments to the flowchart of the Colombian consensus will be communicated as appropriate. Following this meeting, conclusions and recommendations to adjust the flowchart for this population will be created.

Expected results Just in northern of Colombia, there were at least four cases of maternal mortality associated with pathologies in which the diagnosis of a TMA was a differential diagnosis in 2014. By mid-2015, over 4,000 cases of severe maternal morbidity have been documented in Colombia, with diagnoses defined as hypertensive disorder of pregnancy, thus a proposal of an adequate algorithm to discriminate the different TMA in the obstetric patient will minimize underdiagnoses of aHUS.