Transient diabetes insipidus in critically ill COVID19 patients

Abstract

Purpose

Vasopressin has become an important vasopressor drug while treating a critically ill patient to maintain adequate mean arterial pressure. Diabetes insipidus (DI) is a rare syndrome characterized by the excretion of a large volume of diluted urine, inappropriate for water homeostasis. We noticed that several COVID19 patients developed excessive polyuria suggestive of DI, with a concomitant plasma sodium-level increase and/or low urine osmolality. We noticed a temporal relationship between vasopressin treatment cessation and polyuria periods. We reviewed those cases to better describe this phenomenon.

Methods

We retrospectively collected COVID19 ECMO patients' (from July 6, 2020, to November 30, 2021) data from the electronic medical records. By examining urine output, urine osmolality (if applicable), plasma sodium level, and plasma osmolality, we set DI diagnosis. We described the clinical course of DI episodes and compared baseline characteristics between patients who developed DI and those who did not.

Results

Out of 37 patients, 12 had 18 episodes of DI. These patients were 7 years younger and had lower severity scores (APACHE-II and SOFA). Mortality difference was not seen between groups. 17 episodes occurred after vasopressin discontinuation; 14 episodes were treated with vasopressin reinstitution. DI lasted for a median of 21 h, with a median increase of 14 mEq/L of sodium.

Conclusions

Temporary DI prevalence after vasopressin discontinuation in COVID19 ECMO patients might be higher than previously described for vasopressin-treated patients.

1. Introduction

Vasopressin has become an important vasopressor drug while treating a critically ill patient [[1], [2], [3]]. Several works demonstrated vasopressin deficiency in several shock states [4,5]. Therefore, its usual indication is in severe cases of septic shock, in which the target mean arterial pressure is not achieved with fluid resuscitation and noradrenalin treatment; vasopressin deficiency after cardiothoracic surgery [6]; and the need for blood pressure augmentation in cases of brain injury [7]. Blood pressure elevation is mediated through V₁ receptor activation on vascular smooth muscle (mediated both by phospholipase C activation and intracellular Ca⁺² levels increment and by blockade of K^+-sensitive ATP channels that prevents intracellular Ca⁺² to decrease), distinct from the V₂ receptors in the renal collecting ducts (a G-coupled protein, activation of which causes intracellular cyclic adenosine monophosphate increment), which mediate its antidiuretic effect [8].

Diabetes insipidus (DI) is a rare syndrome characterized by the excretion of a large volume of diluted urine, which is abnormal concerning water homeostasis and not caused by any other osmotic diuresis. Therefore, the basic diagnostic criteria are the presence of polyuria (more than 2.5–3 L/day, equivalent to more than 100 mL/h, or more than 40–50 mL/Kg/day), with submaximal concentrated urine (urine osmolality of 300 mOsm/Kg or less) [[9], [10], [11]]. Another definition does not account for polyuria, but rather the presence of hypernatremia with submaximal concentrated urine [12]. These definitions are complementary – i.e., patients with DI can have plasma sodium and osmolarity levels within normal limits with polyuria, or hypernatremia without polyuria. At the usual critical care practice, DI is usually considered in the differential diagnosis when hypernatremia ensues; however, if the patient receives enough fluids to avoid significant negative fluid balance, hypernatremia and hyperosmolarity may not develop.

During the COVID19 pandemic, we noticed that several COVID19 patients who were treated with venous-venous (VV) extracorporal membrane oxygenation (ECMO) developed excessive polyuria, with a concomitant plasma sodium-level increase (not always to hypernatremia levels). As the calculated plasma osmolarity was increased, we considered those episodes as DI, although we were unable to measure urine osmolality and electrolytes in some of these patients. We noticed a temporal relationship between vasopressin treatment cessation and polyuria periods (after vasopressin withdrawal) in almost all the cases. This was described only scarcely.

We describe the course of polyuria with a clinical diagnosis of DI of 12 patients who were admitted to our COVID19 intensive care unit from July 6, 2020, to November 30, 2021. All these patients were treated with VV-ECMO; all but one was treated with vasopressin.

2. Methods

After the Rabin Medical Center gave IRB approval, we retrospectively reviewed the electronic medical charts of all COVID19 ECMO-treated patients from July 6, 2020, to November 30, 2021. We collected demographic data (age, sex, weight, and body mass index [BMI]), severity scores calculated (Acute Physiology And Chronic Health Evaluation II (APACHE-II), Sequential Organ Assessment Failure at 24 h from ICU admission (SOFA24)), and clinical data, including ECMO duration, urine output, vasopressin treatment duration, vasopressin indication, vasopressin dose, diuretic use, sodium levels before, during, and after the polyuria; diuretic use before and during polyuria; measured plasma osmolarity (if it was tested) or calculated plasma osmolarity based on routine chemistry according to the estimated osmolarity formula [13] (2Na(mEq/L) + Urea(mg/dL)/2.8 + Glucose(mg/dL)/18); and urine osmolality and urine sodium levels. We also followed patient outcomes (mortality or survival to hospital discharge). Data regarding daily fluid balance before the polyuria phase (input and output) were collected when applicable. (These data must be interpreted with caution as they were not available in the COVID19 ICU because of software shortage; they were available only after the COVID19 patient was considered noninfectious and was placed in the general ICU. Therefore, the fluid balance was only calculated for the later period.) Glycemic control indices were used based on ward protocol.

The indication for vasopressin use was determined by reviewing the medical charts retrospectively based on the daily follow-up and microbiological laboratory results. We categorized the indications into four groups: (1) sepsis – in case of proven septic shock; (2) suspected sepsis – in cases where the treating physician diagnosed sepsis despite no microbiological proof of infection; (3) ECMO sedation – in cases where hypotension was due to sedation and/or beta-blocker use and no suspicion of sepsis (vasopressin is sometimes preferred over noradrenaline to avoid tachycardia, which might worsen patients oxygenation; sedation is usually with morphine, propofol, midazolam, and ketamine (if needed) to achieve deep sedation to prevent mechanical problems with ECMO cannulas because of patient self-movement); and (4) polyuria/DI – in cases where the daily follow-up indicated that vasopressin was prescribed because of polyuria or DI, without any of the above criteria.

Definite DI diagnosis was set if all of the following criteria occurred concomitantly [[10], [11], [12]]:

• 1.Polyuria – either urine output rate greater than 2.5 L/day or 100 ml/h or 40 mL/Kg/day
• 2.Plasma osmolality not lower than 275 mOsm/Kg
• 3.Urine osmolality less than 300 mOsm/kg

In the cases where urine osmolarity was not applicable, we defined DI as probable or possible based on sodium level, in addition to criteria 1 and 2 mentioned above:

Probable – plasma sodium level equal to or greater than 147 mEq/L [14].

Possible – plasma sodium increase of at least 10 mEq/L before and during the polyuria.

In cases where the patient had polyuria with low plasma osmolality (less than 275 mOsm/Kg) or polyuria that did not meet any of the above criteria, we estimated that the polyuria is due to fluid overload.

Based on the abovementioned definitions, patients were divided into two groups: those who developed DI and those who did not. Naranjo score was calculated for each patient in the DI group considering vasopressin cessation as the offending drug.

Because of the relatively small number of patients (and therefore with assumption of non-normally distributed data), we used median (IQR) to compare baseline characteristics with Mann–Whitney test. Osmolarity and sodium levels of patients were compared between polyuria episodes of patients in the DI group and patients in the non-DI group using Mann–Whitney test.

3. Results

In the period from July 6, 2020, to November 30, 2021, we treated 152 patients in the COVID19 intensive care unit. A total of 37 patients were treated with VV-ECMO. Table 1 compares the baseline characteristics of all ECMO-treated patients.

Table 1.

Baseline characteristics.^⁎

	All ECMO patients
	Patients who developed DI	Patients who did not develop DI
N	12	25
Age (years)	45.31, (38.64, 54.65)	55.26, (46.06, 58.85)	p = 0.066
Male sex (n, %)	6 (50%)	16 (64%)	p = 0.41
BMI (Kg/m2)	27.7, (25.9, 33.2)	34.11, (25.09, 40.84)	p = 0.022
APACHE-II	11.5, (9.5, 13.75)	15, (13, 19)	p = 0.033
SOFA24	4.5, (3, 8)	7, (4,9)	p = 0.12
Vasopressin-treated patients (n, %)	11 (91.67%)	20 (80%)	p = 0.367
Length of stay (days)	13.41, (11.44, 16.72)	17.1, (11.76, 21)	p = 0.327
Hospital mortality (n, %)	6 (50%)	19 (76%)	p = 0.114

Numerical variables are presented as median (IQR).

^⁎p-value is based on Mann–Whitney test for numerical variables and chi-square test for categoric variables.

We found 35 episodes of polyuria: 18 polyuria episodes in 12 patients were diagnosed as DI (whether definite/probable/possible DI based on the abovementioned criteria). Eight patients had one episode, three patients had two episodes, and one patient had four episodes. Table 2 details the baseline characteristics of the 12 patients who had DI. Three patients were pregnant upon ICU admission (they were all delivered shortly after ICU admission and connected to ECMO afterward). Five patients were referrals from non-ECMO-capable hospitals. Supplement 1 provides more detailed information about those patients.

Table 2.

Baseline characteristics of the 12 patients who developed DI.

Patient #	sex	Age (years)	Weight (Kg)	BMI (Kg/m2)	Covid19 Vaccination status	Pregnancy	APACHE-II	SOFA24	Naranjo Score	Days from COVID19 diagnosis to ventilation	Days from ventilation to ECMO	Days on ECMO	Days from COVID19 diagnosis to
													Mortality	Discharge
1	F	33.5	65	23.9	No	Yes	8	8	6	8	13	25		63
2	F	49.6	105	36.3	No	No	13	8	4	8	9	27	44
3	F	40.3	65	27.1	No	Yes	10	3	4	13	0	13		28
4	F	60.1	83	32.4	No	No	20	8	4	9	6	24	39
5	M	45.5	94	34.5	No	N/A	10	4	6	21	0	66	87
6	M	54.9	80	24.7	No	N/A	16	11	6	7	0	35	42
7⁎	F	54.6	85	29.4	Yes	No	24	5	0	11	0	10		29
8	M	60.8	N/A	N/A	No	N/A	13	5	4	17	1	5	23
9	M	50	70	22.1	No	N/A	7	3	4	10	0	19	29
10	M	36	140	43.2	No	N/A	8	4	4	5	10	36	51
11	M	26.7	80	27.7	No	N/A	6	2	4	13	8	24		69
12	F	24.7	75	27.5	No	Yes	12	3	6	15	0	46		74

^⁎Patient is immunocompromised – treated with rituximab due to myasthenia gravis; this patient was not treated with vasopressin.

Naranjo score was calculated for each patient as if vasopressin cessation is the offending drug. The median Naranjo score was 4 (IQR 4–6).

Tables 3a and 3b present the clinical course of the DI episodes; some of the patients had more than one DI episode as efforts to withhold vasopressin treatment were made (and failed). Diuretics were in use in only one episode; one episode was in relation to septic episode with non-oliguric acute kidney injury (AKI) treated with hemadsorption using continuous renal replacement therapy (CRRT).

Table 3a.

Clinical course of DI episodes – vasopressin and polyuria characteristics.

Patient #	Polyuria episode #	Days from ECMO start to vasopressin treatment	Days from ECMO start to vasopressin cessation	Vasopressin indication	Hours from vasopressin cessation until polyuria	Polyuria duration (hours)	Polyuria volume (ml)	Urine output during polyuria		Urine output after polyuria stopped		Vasopressin duration after polyuria (hours)	DI diagnosis
								ml/h	ml/Kg/d	ml/h	ml/kg/day
1	1	0.42	11.4	Susp. Sepsis	1	24	9170	382	141	133	49	46	Definite
	4a	15.1	21.5	DI	2	27	7450	276	102	75	28	27	Possible
2	1a	0	2	Sepsis	49	45	10,230	227	52	56	13	48	Definite
3	1b	1	2	ECMO sedation	0	48	5400	113	42	77	28	0	Probable
4	1	11	14.6	ECMO sedation	3	11	2340	213	62	87	25	126	Definite
5	1	5.6	8.5	ECMO sedation	4	29	8450	291	74	142	36	104	Possible
	2	9.5	13.9		5	8	6000	750	191	113	29	12	Probable
	3a	14.3	14.8		5	10	1720	172	44	90	23	76	Probable
	4	16.9	20.1		9	21	5480	261	67	113	29	49	Definite
6	1	0	1	Sepsis	30	24	3770	157	47	83	25	161.4	Probable
	6	11.5	16.5	Susp. sepsis	68	14	2550	182	55	80	24	375	Probable
7	1	N/A	N/A	N/A	N/A	12	1200	100	28	23	7	0	Definite
8	1	-1	2	ECMO sedation	15.8	23	5310	231	69	47	14	48.9	Possible
9	1	4	6	Susp. Sepsis	10	20	3150	158	54	43	15	27.7	Definite
10	1c	1.7	8.9	Sepsis	12	19	6750	355	61	100	17	5	Definite
11	1	1	6.3	ECMO sedation	2	15	3160	211	63	50	15	11	Definite
12	1	9.8	12.1	ECMO sedation	3	21	5465	260	83	57	18	0	Definite
	2b	26.8	27.3	Sepsis	4	71	10,910	154	49	62	20	0	Definite

^aPatient was treated with DDAVP to diminish urine output; without improvement, vasopressin was restarted.

^bPatient was treated with enteral water; vasopressin was not prescribed; polyuria resolved spontaneously.

^cPatient was treated with 10 mg bolus of furosemide during the polyuria.

^dPatient was treated with CRRT during this episode as part of hemadsorption treatment due to sepsis. CRRT with a hemadsorption filter was connected to the ECMO, prescribed with an effluent dose of 18.6 ml/h/kg and filtration fraction of 18.7%. Hemadsorption was stopped because of filter clotting 60 h from its onset. Patient creatinine declined from 0.6 mg/dL to baseline of 0.3 mg/dL. The patient was non-oliguric during all CRRT duration. A total of 3000 ml were removed during treatment.

Table 3b.

Clinical course of DI episodes – laboratory features.

Patient #	Polyuria episode #	Polyuria volume (ml)	Plasma osmolarity during polyuria (mOsm/Kg)	Urine osmolarity during polyuria (mOsm/Kg)	Urinary sodium during polyuria (mEq/L)	Last sodium level before polyuria (mEq/L)	Sodium level during polyuria (mEq/L)	Sodium level after polyuria stopped (mEq/L)	DI diagnosis
1	1	9170	299	125	35	133	151	138	Definite
	4a	7450	280⁎	N/A	N/A	125	140	136	Possible
2	1a	10,230	291⁎	71	20	139	139	140	Definite
3	1b	5400	312⁎	N/A	76	138	155	142	Probable
4	1	2340	286⁎	136	13	132	137	138	Definite
5	1	8450	299⁎	N/A	N/A	123	142	139	Possible
	2	6000	300⁎	N/A	N/A	126	147	146	Probable
	3a	1720#	303⁎	N/A	N/A	145	147	141	Probable
	4	5480#	297⁎	100	N/A	129	145	142	Definite
6	1	3770	327⁎	N/A	N/A	139	148	146	Probable
	6	2550	299⁎	N/A	N/A	144	148	145	Probable
7	1	1200	297	200	72	139	147	145	Definite
8	1	5310	293	N/A	N/A	120	141	136	Possible
9	1	3150	338	281	36	132	149	140	Definite
10	1c	6750	303⁎	61	0	127	146	143	Definite
11	1	3160	296	155	N/A	123	140	143	Definite
12	1	5465	302	64	N/A	130	149	144	Definite
	2d	10,910	297	136	26	140	142	140	Definite

^⁎Plasma osmolarity is calculated based on measured sodium, glucose, and urea levels based on the plasma osmolarity calculation equation.

^aPatient was treated with DDAVP to diminish urine output; without improvement, vasopressin was restarted.

^bPatient was treated with enteral water; vasopressin was not prescribed; Polyuria resolved spontaneously.

^cPatient was treated with 10 mg bolus of furosemide during the polyuria.

^dPatient was treated with CRRT during this episode as part of hemadsorption treatment due to sepsis. CRRT with a hemadsorption filter was connected to the ECMO, prescribed with an effluent dose of 18.6 ml/h/kg and filtration fraction of 18.7%. Hemadsorption was stopped because of filter clotting 60 h from its onset. Patient creatinine declined from 0.6 mg/dL to baseline of 0.3 mg/dL. The patient was non-oliguric during all CRRT duration. A total of 3000 ml were removed during treatment.

Vasopressin was given to 11 of 12 patients who developed polyuria at a dose range of 1–2 IU/h. Vasopressin was given to 6 patients because of sepsis (proven or suspected); the other 5 patients received vasopressin for hypotension caused by sedation and/or beta-blocker treatment. The median time of vasopressin treatment was 3 days (IQR 2-5d). All DI episodes (but one) were after vasopressin discontinuation, with a median time of 5 h from vasopressin discontinuation until DI (IQR 3 − 12 h). 14 episodes of DI were treated with reinstitution of vasopressin to diminish urine output (in half of the episodes at a dose of 1 IU/h and in the other half at a lower dose of 0.5 IU/h) until it could be stopped without polyuria recurrency (usually without downward titration); the other 4 episodes resolved spontaneously. Significant differences in sodium levels were found before, during, and after the DI. Table 4 further describes the DI episodes.

Table 4.

Description of polyuria episodes.

	DI episodes	Non-DI polyuria episodes	P value
Polyuria duration (hours)	21 (14.25–26.25)	14 (10.25–17.75)	p = 0.03
Polyuria volume (ml)	5432 (3152–7275)	2605 (1942–4167)	p = 0.006
Plasma osmolarity during polyuria (mOsm/Kg)	299 (296.25–303.75)	285.5 (296.25–302.75)	P < 0.001
Sodium level (mEq/L)
Pre-polyuria	132 (126.25–139)⁎	130.5 (124.5–136.5)^	p = 0.52
During polyuria	146.5 (141.25–148)⁎	137 (130.25–140.75)^	p < 0.001
Post-polyuria	141.5 (149.25–143.75)⁎	135 (131.5–138)^	p < 0.001
Duration of vasopressin treatment after polyuria onset, until complete cessation (hours)	36.85 (6.5–69.25)	42 (18–74)	p = 0.7

Median levels(IQR)are presented.

P value is calculated using the Mann–Whitney U test.

^p > 0.05 between sodium levels in the pre/during/post-polyuria in the non-DI group.

^⁎p < 0.01 between sodium levels in the pre/during/post-polyuria in the DI group.

Supplements 2a and 2b list all polyuria episodes, including those that were judged to be a result of fluid overload (and not DI), whether because of low plasma osmolarity, low sodium level, or both. They also include the data regarding daily fluid balance before the polyuria phase. It should be interpreted with caution as the data were not available for most of the period.

Interestingly, although no significant difference was noted between the DI polyuria episodes and non-DI polyuria episodes in pre-polyuria median sodium level, a significantly higher median sodium level was found in the DI episodes during and after the polyuria; plasma osmolarity was also significantly higher in the DI episodes. See Table 4 for further details.

4. Discussion

The first reports regarding the possible effects of vasopressin on water hemostasis relate to hyponatremia that developed during vasopressin infusion [15,16]. In 1995, Melo et al. [16] described a patient who underwent a cesarian section because of upper GI bleeding and was treated with vasopressin due to variceal bleeding. Vasopressin was stopped because of hyponatremia, and then the patient had diuresis with an increase in sodium plasma level along with low urine osmolality. In 2004, Kristeller and Sterns [17] were the first to report the term transient DI (tDI) after vasopressin discontinuation. They described a patient with a neurosurgical background and SIADH, who temporarily had polyuria and hypernatremia after vasopressin discontinuation.

Until 2019 and the COVID19 pandemic, several case reports and two case series described 47 patients who developed a clinical course of DI after vasopressin discontinuation. Not all reports include urine osmolarity results, but all reports described patients with polyuria and increased plasma sodium level (or its increment during polyuria) [7,14,[18], [19], [20], [21], [22], [23], [24], [25]]. Of these 47 patients, only 6 were treated with ECMO. In 37 cases, the diagnosis of DI was made without urine osmolarity evaluation (in six cases, urine-specific gravity was measured as an indicator for diluted urine) [14,19,20].

Regarding ECMO-related DI, we could find another four reports since 2019, but only two of them were treated with vasopressin while on VV-ECMO treatment and developed DI upon vasopressin discontinuation [[26], [27], [28], [29]].

Recently, during the COVID19 pandemic, another four relevant cases were published. Two describe critical COVID19 patients treated with VV-ECMO who were diagnosed with DI. Only one of these cases is similar to our description – a VV-ECMO patient who was weaned from vasopressin a day before polyuria appeared [30]. The other descriptions are of a patient who was not treated with vasopressin [31,32]; a non-ECMO COVID-19 patient who developed central DI with brain edema [33]; and an ECMO patient who developed oliguria and hyponatremia while treated with vasopressin [34]. There are also three case reports of DI development as late sequala of mild to moderate COVID19 infection (4–8 weeks after COVID19 infection) [[35], [36], [37]]. Supplement 3 provides a more detailed literature review.

Over a 10-month period, we evaluated 12 patients with polyuria, of which 9 had a definite diagnosis of DI, 2 of probable DI, and 1 of possible DI.

All 12 patients were diagnosed with COVID19 and treated with VV-ECMO because of COVID19 hypoxemic failure. All but one patient was treated with vasopressin. The duration of vasopressin treatment and the time between vasopressin discontinuation and polyuria appearance were similar to the previously reported series [7,14]. All these patients had polyuria, and plasma osmolarity was within normal limits. As we could not examine urine osmolality in all these patients, we suggest the terms definite, probable, and possible DI. All 18 polyuria episodes presented here can be counted as DI based on these definitions. It is worth emphasizing that hypernatremia develops with DI only in the adipsic patient, while the patients described here received nutrition (whether enteral or parenteral), which might explain the absence of hypernatremia (or sodium increment absence) in some of the patients. This emphasizes the basic principle that was set in the abovementioned DI criteria: DI diagnosis can be set in case of a large volume of diluted urine, which is inappropriate to the volume status of the patient (not hypervolemic, i.e., plasma osmolarity in not low). With this definition, plasma sodium and plasma osmolarity may be within normal limits or may be increased (which is supportive of DI diagnosis).

The major forms of DI are central DI caused by hypothalamic–posterior pituitary axis dysfunction with decreased or absent antidiuretic hormone (ADH) secretion and nephrogenic DI, resulting from kidney insensitivity to ADH. Another form is gestational DI, which is caused by ADH degradation by placental-produced vasopressinase. As polyuria resolution occurred with vasopressin reinstitution in 10 of the episodes presented here (before vasopressin could be eventually stopped), it cannot be classified as nephrogenic DI; it can be allegedly classified as central DI. However, none of the patients had a known CNS pathology. As all patients were treated with ECMO, we could not perform an MRI (ECMO equipment is not MRI-compatible). A brain CT was performed on one patient (patient #9), without demonstrating any pathology. None of the survivors demonstrated neurological symptoms after weaning from sedation.

As all patients were eventually weaned from vasopressin without polyuria recurrency, it seems reasonable to classify those episodes as tDI.

There are several possible explanations for the development of tDI during or after vasopressin treatment. The first is V₂ receptors downregulation during supraphysiological vasopressin treatment. During vasopressin treatment, there is a temporal deficiency in V₂ receptor activation. A previous rat study showed that the V₂ receptor downregulation appears from the first day of DDAVP treatment until its maximal effect on the fifth day of treatment. This effect is over only after V₂ receptors are upregulated, which takes some time, in which there is an apparent (and transient) DI. Plasma osmolarity and plasma sodium levels during the DI depend on the volume and electrolyte status at the onset of DI. As V₂ receptors are not completely absent, reinstitution of vasopressin maintains antidiuretic effects. Thus, it is likely that tDI cannot be regarded as central DI (even though it improves with vasopressin treatment) nor as nephrogenic DI [7,14,38].

A second explanation is decreased secretion of vasopressin from the pituitary (whether because of decreased production, depleted stores, or secretion defect), a mechanism that might be exaggerated by steroid treatment [39]. This mechanism is better understood in septic shock, a situation where a low endogenous level of vasopressin has been demonstrated [4,5]. About half of the patients in this series were treated with vasopressin due to sepsis (proven or suspected). The other half were treated with vasopressin due to sedation.

There are several reports of neurological manifestations of COVID19. The suggested mechanisms are a direct invasion of the virus to CNS neurons (including hypophysial cells) and delayed immune response targeting the CNS (some of the data relate to SARS-CoV and not directly to SARS-CoV2) [[40], [41], [42]]. It is possible that the COIVD19 infection (directly or mediated by inflammatory state or immune response) causes decreased vasopressin secretion from the pituitary, which might be exaggerated by exogenous vasopressin treatment. This might be the reason for the high rates of tDI we encountered in this patient population. The pathophysiological effects of COVID19 might explain why we noted tDI in an ECMO patient who was not treated with vasopressin. As the neurological effects of COVID19 are somewhat delayed, this might explain why the resolution of tDI was prolonged in some of the patients.

Another explanation is a variation in vasopressinase activity coded by several genotypes, one of which is associated with increased vasopressin clearance [43]. Regarding vasopressinase activity, it is possible that the DI in the three pregnant patients was gestational DI as this might appear up to 6 weeks after delivery [10].

The series presented here differs from the other cases of tDI in several aspects. We encountered tDI during vasopressin discontinuation in 11 ECMO patients (out of 37 ECMO patients; 29.73%) in 10 months, whereas in literature review 51 patients were described over 25 years. Clinicians should be alert to the possibility of tDI, which may be more prevalent than previously described in vasopressin-treated patients [14]. Second, all were COVID19 patients, who were treated with ECMO. To the best of our knowledge, this is the first series that describes tDI both in COVID19 patients and VV-ECMO patients (description of tDI in non-COVID19 VV-ECMO patients was also very scarce). We could find only two case reports in critically ill COIVD19 patients who were treated with ECMO and another one a non-ECMO COVID19 patient. We can assume that some of the higher prevalence we found is related to the COVID19–ECMO combination. It is possible that the mechanism that predisposed these patients to tDI caused the one patient who was not treated with vasopressin to develop this clinical course. Third, despite the COVID19 burden, we were able to prove DI with urine osmolarity measurement in 9 out of 12 patients. To date, this is the largest series regarding tDI, which includes this data.

Because of the retrospective nature of this series and the pandemic nature of the COIVD19 ICU platform, it has some limitations. First, the exact fluid balance (including IV fluid treatment) is not known. The COVID19 ICU was not equipped with the necessary software and hardware to calculate daily and cumulative fluid balance. It is, therefore, theoretically possible that the abovementioned tDI episodes reflected only fluid overload diuresis. However, it seems very unlikely with low urine osmolarity and high plasma osmolarity. Second, because of COVID19 ICU software shortage, we cannot account for sodium-level changes caused by other medications or dilutional fluids of IV drugs. Although the default dilutional fluid for IV drugs in our department is glucose 5%, we cannot examine the amount and type of dilutional fluids that were actually in use. Third, we were not able to perform more advanced methods to diagnose DI, such as the measurement of ADH level, copeptin, or vasopressinase activity genotypes. Therefore, we cannot come to a conclusion regarding the tDI etiology.

5. Conclusion

We described a case series of 12 critically ill COVID19 patients treated with ECMO who suffered from tDI; in 11 patients, vasopressin was stopped shortly before tDI appeared. This has not been described yet with ECMO and/or COVID19. Clinicians should be aware of this possible adverse effect when weaning from vasopressin. Further research is needed to better describe this phenomenon.

List of abbreviations

COVID19	Corona Virus Disease 2019
ECMO	extracorporeal membrane oxygenation
DI	diabetes insipidus
tDI	temporary diabetes insipidus
APACHE-II	acute physiology and chronic health evaluation II
SOFA	sequential organ failure assessment
ADH - antidiuretic hormone
BMI	body mass index
SIADH	syndrome of inappropriate ADH secretion

Funding

No funding was received for this work.

CRediT authorship contribution statement

Liran Statlender: Conceptualization, Data curation, Formal analysis, Writing – original draft. Guy Fishman: Conceptualization, Writing – review & editing, Validation. Moran Hellerman: Data curation. Ilya Kagan: Data curation. Itai Bendavid: Writing – review & editing. Dan Gorfil: Data curation, Writing – original draft, Writing – review & editing. Shani Kaptzon: Data curation. Pierre Singer: Writing – review & editing, Supervision.

Declaration of Competing Interest

There are no conflicts of interest.