*3.3. Predictors of Superimposed Preeclampsia*

To assess variables associated with preeclampsia in CKD patients, Cox regression analysis with stepwise backward selection was performed (Table 2). Variables with *p* < 0.10 at the group comparison were included in the final model.

**Table 2.** Univariate and multivariate Cox regression analysis of predictors associated with superimposed preeclampsia on chronic kidney disease (CKD).


Cox multivariate regression with backward stepwise selection (variables introduced in the first step: maternal age, nulliparity, creatinine at referral, proteinuria ≥1 g/day at referral, pre-existing hypertension; variables that remained in the last step: proteinuria ≥1 g/day at referral, pre-existing hypertension); HR—hazard ratio; CI—confidence interval.

By univariate Cox analysis, proteinuria ≥ 1 g/day at referral was the only statistically significant variable as a risk factor for preeclampsia (*p* = 0.008), followed by pre-existing hypertension, which presented a tendency of significance (*p* = 0.07). In multivariate Cox analysis, proteinuria ≥1 g/day at referral and pre-existing hypertension after adjustment for age, nulliparity, and creatinine, were independent predictors of preeclampsia. The presence of proteinuria ≥1 g/day at referral and pre-existing hypertension increased the risk of preeclampsia 4.1 times (adjusted HR = 4.10, 95%CI: 1.52–11.02, *p* = 0.005) and 2.6 times (adjusted HR = 2.62, 95%CI: 1.01-6.77, *p* = 0.04), respectively.

In Figure 1A,B, the Kaplan–Meier curves show that the cumulative risk of preeclampsia at 35 and 38 weeks of gestation was significantly higher in pregnant women with proteinuria ≥1 g/day (68% and 81%, respectively) than in those with proteinuria <1 g/day at referral (27% and 34%, respectively) (log-rank, *p* = 0.003) and the cumulative risk of preeclampsia was also higher in pregnant women with pre-existing hypertension (67% and 70%, respectively) than is those without pre-existing hypertension (30% and 43%, respectively) (log-rank, *p* = 0.05).

**Figure 1.** (**A**) Kaplan–Meier curves showing the cumulative risk of preeclampsia in pregnant women with proteinuria ≥1 g/day vs. <1 g/day at referral; (**B**) Kaplan–Meier curves showing the cumulative risk of preeclampsia in pregnant women with vs. without pre-existing hypertension.

We evaluated the predictive ability of the identified risk factors and found that proteinuria ≥1 g/day at referral and pre-existing hypertension could predict preeclampsia in pregnant women with CKD with an accuracy of 73.53% (95%CI: 55.6%–87.1%) and 64.71% (95%CI: 46.5%–80.3%), respectively (Table 3).


**Table 3.** Predictive analysis of risk factors for preeclampsia.

CI—confidence interval; LHR—likelihood ratio; PPV—positive predictive value; NPV—negative predictive value.

#### **4. Discussion**

Results from our study showed that pregnant women with CKD are at high risk of developing preeclampsia. Superimposed preeclampsia on CKD complicated 55.8% of the pregnancies in our study. This result is in agreement with other studies that reported values between 21% and 79% [6]. Moreover, the frequency of preeclampsia was increased in CKD stages 3–4. The increased incidence of preeclampsia compared to other studies could have been influenced by the diagnostic criteria we used, according to ACOG (2013) and the revised statement of ISSHP (2014), which exclude proteinuria as a mandatory condition for the diagnosis of preeclampsia [9,10]. The mechanisms by which CKD increases the risk of preeclampsia in not completely understood. Some mechanistical processes have been proposed, such as renin-angiotensin-aldosterone system (RAAS) dysregulation, endothelial dysfunction and complement dysregulation [12–14]. We reported no cases of postpartum preeclampsia within the first 6 months and no maternal deaths. However, one fetal death (5.3%) occurred in the preeclampsia group.

We evaluated possible clinical and biological predictors of superimposed preeclampsia in pregnant women with CKD and we found that the presence of proteinuria ≥1 g/day at referral and pre-existing hypertension increased the risk of superimposed preeclampsia 4.1 and 2.6 times, respectively. Other classic risk factors such as nulliparity, maternal age, BMI and the primary cause of CKD were not significantly associated with preeclampsia.

Proteinuria ≥1 g/day at referral was associated with a significantly increased cumulative risk of preeclampsia of 68% and 81% at 35 and 38 weeks respectively. Similar to our results, data from a prospective study, which included 504 pregnant women with CKD, showed that baseline proteinuria ≥1 g/day was an independent risk factor for pregnancy outcomes [15]. Also, data from a cohort study indicated that the proteinuria level at the beginning of pregnancy was strongly associated with subsequent preeclampsia in pregnant women with CKD [16]. A systemic review and meta-analysis which included 23 studies that evaluated the outcomes of pregnancy in CKD and CKD outcomes in pregnancy, demonstrated that proteinuria >0.5 g/day in pregnant women with CKD was a risk factor for superimposed preeclampsia [17]. Evaluation of proteinuria threshold at the time of referral in our study, until 20 weeks of gestation, is an expression of CKD and we proposed this clinical factor as a predictor and not as a diagnostic marker for superimposed preeclampsia.

According to our analysis, 63.2% of patients with preeclampsia had pre-existing hypertension. In agreement with our results, Braham et al reported that chronic hypertension increased the risk of superimposed preeclampsia 7.7 times in the US population and that the estimated incidence of disease was between 21% and 31.5% [5]. Another study by Webster et al. showed that the incidence of

preeclampsia in women with chronic hypertension was 21% [18]. The higher percentage of preeclampsia in this subgroup could be explained by the fact that all these patients also had CKD and that a possible intrarenal renin-angiotensin-aldosterone system (RAAS) could be involved. It has been shown that in preeclamptic women, systemic RAAS components have lower circulating levels and there is an increased sensitivity to angiotensin II action, but little is known about the role of intrarenal RAAS [19,20]. Angiotensin II present in the renal tissue could be provided by angiotensinogen action, synthesized locally from tubular epithelial cells [21]. Furthermore, it has been demonstrated that angiotensin II promotes sFlt1 production in proximal tubular cells [22]. Yilmaz et al demonstrated that elevated intrarenal RAAS activity is involved in the development of preeclampsia, using elevated urinary angiotensinogen as a marker of local RAAS activation [23].

The diagnosis of superimposed preeclampsia on CKD is difficult to establish in clinical practice because of the common features shared by CKD and preeclampsia, thus making other parameters necessary for diagnosis. Promising results from previous studies showed that angiogenic (placenta growth factor (PIGF)) and antiangiogenic (soluble fms-like tyrosine kanise-1 (sFlt1)) factors and uteroplacental flows could be used as diagnostic tests for superimposed preeclampsia on CKD. However, they have not been implemented in clinical practice, making further study necessary [24–26]. More recently, plasma and urinary biomarkers derived from endothelial tissue (hyaluronan, intercellular adhesion molecule, vascular cell adhesion molecule, P-selectin, E-selectin), biomarkers of RASS and complement system activation have been studied in pregnant women with CKD. Kate Wiles et al. showed that increased plasma concentration of hyaluronan and vascular cell adhesion molecule had the potential to differentiate between pregnant women with superimposed preeclampsia on CKD and those without superimposed preeclampsia and that endothelial dysfunction had an important role in the pathophysiology of superimposed preeclampsia [13]. Relying on the latest evidence, the diagnostic utility of angiogenic, antiangiogenic and endothelial derived biomarkers in superimposed preeclampsia in CKD pregnant women is recognizable, but they are limited by availability and costs in clinical practice. Therefore, classical factors such as level of proteinuria at referral or pre-existing hypertension may still be useful in assessing the risk for superimposed preeclampsia on CKD [27] and could also contribute to the development of various prognostic tools alongside newly discovered biomarkers.

The strengths of our study are the use of adjusted models in the multivariate analysis and the diagnostic criteria used. However, its limitations such as the retrospective design, small size of the cohort study, and its single-center nature may limit its generalizability. Another limitation of our study that must be mentioned is regarding the selection bias. All patients included in the study had CKD and this could explain the higher incidence of PE than what was reported in other studies.

## **5. Conclusions**

In conclusion, findings from our study indicate that pregnant women with pre-existing chronic kidney disease are at high risk of developing preeclampsia and that proteinuria ≥1 g/day at referral and pre-existing hypertension are independent predictors of superimposed preeclampsia. Consequently, maintaining proteinuria of <1 g/day in women with CKD who want to become pregnant could decrease the risk of preeclampsia. Moreover, women with pre-existing hypertension should be intensively monitored during pregnancy. However, to translate this conclusion into clinical practice is difficult considering the small size of our study population. Thus, larger studies are needed.

**Author Contributions:** Conceptualization, B.M.S, G.I., C.B and M.H.; Data curation, A.A., R.J. and B.O.; Formal analysis, B.M.S.; Investigation, A.A., R.J. and B.O; Methodology, B.M.S., G.I., C.B. and M.H.; Project administration, B.M.S.; Resources, A.A., R.J. and B.O.; Software, B.M.S.; Validation, A.A., R.J. and B.O.; Visualization, A.A.; Writing—original draft, B.M.S., G.I., C.B. and M.H.; Writing—review and editing, B.M.S., A.A., G.I., R.J., B.O., C.B. and M.H. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.
