A minimum of two years of undertreated primary hypothyroidism, as a result of drug-induced malabsorption of l-thyroxine, may have metabolic and cardiovascular consequences

Highlights • L-T4 malabsorption is frequently encountered in clinical practice.• Drug induced L-T4 malabsorption has metabolic and cardiovascular consequences.• Control of TSH is not enough when drug induced L-T4 malabsorption is suspected.


Introduction
Oral levothyroxine (L-T4) is the recommended hormone therapy for hypothyroidism, a therapy that is monitored by periodic measurements of serum TSH levels [1,2] when the hypothyroidism etiology is primary.
Undertreated primary hypothyroidism, that is elevation of serum TSH above target levels, is observed in approximately 20% of primary hypothyroid patients treated with tablet L-T4 [3,4]. In approximately onethird of such patients, the cause of undertreatment is oral ingestion of one or more medications that interfere with the intestinal absorption of https://doi.org/10.1016/j.jcte.2019.100189 L-T4 [3]. Several lists of such interfering drugs are available (see Patients and Methods). The recommended serum TSH target level is ≤4.12 mU/L [1][2][3], which is the upper normal limit recorded by the Third National Health and Nutrition Examination Survey (NHANES III) [5]. However, for the National Academy of Clinical Biochemistry [6], which included some thyroid experts, and for other thyroid specialists (for instance, Ref. [7]), the upper normal limit of TSH should be 2.50 mU/L.
The issue of consequences from undertreated hypothyroidism is typically viewed in terms of persistent or partially corrected hypothyroidism symptomatology being associated with poor quality of life [8]. The 2014 ATA guidelines [2] devote one recommendation to the issue ["4d. What are the potential deleterious effects of inadequate levothyroxine?"] upon discussing the literature in 30 lines of text. The strong, moderate quality evidence recommendation is "The adverse effects of thyroid hormone deficiency include detrimental effects on the serum lipid profile and progression of cardiovascular disease. We recommend that patients with overt hypothyroidism be treated with doses of levothyroxine that are adequate to normalize serum thyrotropin levels, in order to reduce or eliminate these undesirable effects" [3,4]. However, serum TSH levels in the upper part or above the upper limit of the reference range are associated with greater rates of metabolic disorders and cardiovascular disease, because of increased levels of fasting blood glucose, total cholesterol, LDL cholesterol, triglycerides, increased frequency of metabolic syndrome or type 2 diabetes mellitus [9][10][11][12][13][14][15][16][17].
Based on the above, in L-T4 treated primary hypothyroid patients who were coingesting tablet L-T4 and at least one interfering medicine, we aimed to assess the magnitude of increase in serum TSH and the occurrence of complications (namely, aggravation of preexisting or de novo appearance of metabolic and cardiovascular events).

Material and methods
This is a retrospective study based on the analysis of computerized records of persons under the care of 8 family physicians; these physicians were previously involved in research studies [18][19][20]. The coauthors of this paper other than the 8 family physicians were blind to the identity of the persons.

Patients
The flow-chart of the study is illustrated in Fig. 1. First, in the 10,426 persons aged ≥18 years under the care of the 8 family physicians, we searched for those who were taking tablet L-T4 therapy, and retrieved 730 persons. The fundamental reasons for L-T4 therapy were autoimmune hypothyroidism and iatrogenic hypothyroidism (post-thyroidectomy almost always).
Third, among these 391 patients we searched those who fulfilled our enrolment criteria for forming the final cohort, and retrieved 114 patients (29.1% of 391, and 15.6% of 730). These patients consisted of 100 women (88%) and 14 men (12%), and their mean age was 65.6 ± 12.7 years. Criteria for enrolment were: (i.) uninterrupted L-T4 therapy with tablet L-T4; (ii.) taking none of the said medications for at least 24 consecutive months in order to provide data for the period of time that we will refer to as "OFF interfering drug(s)" or non-exposure or baseline; (iii) taking any of the said medications for at least another 24 consecutive months in order to provide data for the period of time that we will refer to as "ON interfering drug(s)" or exposure; (iv.) having at least two non-exposure and at least two exposure determinations of serum TSH, all coming from the same laboratory in a given patient; (v.) having clinical and pertinent biochemical data associated with complications (see below) that were recorded within a period of at least two years during which serum TSH had remained on target while OFF interfering drug(s) and within a minimum of another two years while ON interfering drug(s). This criterion was important to compare the proportions of complications (exposure versus non-exposure to interfering drugs).
Duration of the non-exposure and the exposure periods, and definition of complications The databases of all 8 physicians record the date and number of prescriptions of any medicines, thus permitting to calculate the duration of treatment. The first day of ingestion of L-T4 or any interfering drug was the day after the date of the prescriptions. The last day of ingestion of L-T4 or the interfering drug(s) was either explicitly notated (for instance, because of adverse events) or deduced knowing the number of tablets/capsules in each prescribed package and the daily dose.
Concerning complications measurable biochemically, a precaution was taken not to compare the same biochemical index at very different ages. Once we determined the duration of the exposure period for each of the 114 patients, we opted to use a chronological window of similar duration to have the equivalent data for the non-exposure period (baseline).

Outcomes
These were continuous variables and categorical variables. The first consisted of mean ± SD of serum TSH. The second consisted of proportions of serum TSH above two different threshold levels (4.12 or 2.50 mU/L). Another categorical variable was proportions of de novo appearance or aggravation of any of the abnormalities (complications) mentioned above.

Statistics
Differences between continuous variables were analysed by log 10transformed ANOVA, due to nongaussian distribution of serum TSH. Differences between categorical variables by the χ 2 test or by Fisher's exact test, as appropriate. Regardless of test, P values < 0.05 were considered statistically significant.

Table 3
Serum TSH and proportions of TSH levels above two threshold levels (4.12 or 2.50 mU/L) in 114 patients with primary hypothyroidism who are stratified based on two periods of time, namely when they took L-T4 alone (OFF or non-exposure) and when they added ingestion of one or more drugs that interferes with the intestinal absorption of L-T4 (ON or exposure). .

Medications interfering with L-T4 intestinal absorption
The spectrum of such medications in the 114 patients is summarized in Table 2  This comparison is shown in Table 3, where TSH is considered both as a continuous variable and a categorical variable based on two thresholds (see Introduction, and Patients and Methods).
Co-ingestion of tablet L-T4 and any interfering drug(s) was associated with serum TSH levels approximately 100% greater than those associated with ingestion of tablet L-T4 alone, the difference being extremely significant (P = 2.2 × 10 −20 ). * P values typed boldface are statistically significant, while P values typed boldface italics are borderline statistically significant (P between 0.10 and 0.05).  Table 1 and related footnotes. For details of complications in the 2 patients who had at least one TSH assay above the higher threshold (4.12 mU/L), see Table 5

Second outcome: complications during exposure vs non-exposure (baseline)
Overall, aggravation of pre-existing relevant condition(s) or de novo appearance of it/them occurred in two-thirds of patients (76/114) ( Table 4). Mean levels of TSH were 3-fold greater during exposure compared to non-exposure (P = 3.1 × 10 −34 ) in the 76 patients who experienced complications, while they were insignificantly greater in the 38 patients who did not. Proportions of TSH measurements above either threshold were consistently greater during exposure compared to non-exposure. Such proportions were significantly greater in either above TSH threshold category (> 4.12 or > 2.5 mU/L) ( Table 4, footnote).

Relationship of complications with number of TSH measurements above thresholds
Based on a 3-tier categorization (one, two or ≥three TSH measurements above 4.12 mU/L during exposure), the 29 patients in the complication group were distributed significantly differently from the 12 patients in the noncomplication group (P = 0.007) ( Table 5). Finally, in the 86 measurements from the 29 complication patients, serum TSH was significantly greater than in the 41 measurements from the 12 noncomplication patients (5.88 ± 5.68 mU/L [median 4.66] vs 2.54 ± 2.98 [median 0.96], P = 2.9 × 10 −6 ) ( Table 5, footnote). The whole pattern held upon using the threshold of 2.5 mU/L (Table 5).
Noteworthy, 5/114 patients had no pre-existing abnormality during non-exposure and none appeared de novo during exposure. The corresponding TSH levels were 0.85 ± 0.43 (median 0.91) and 1.15 ± 0.59 mU/L (median 1.30), with none of a total of 31 TSH measurements being greater than 2.0 mU/L (data not shown).

Relationship of number of complications with TSH measurements above thresholds
Details on each complication in the whole group of the 76 patients and the two subgroups resulting from the target TSH-based stratification (4.12 mU/L threshold) are shown in Table 6. The two subgroups are graphically contrasted in Figs. 2 and 3. By definition, during exposure no preexisting condition worsened or appeared de novo (if not pre-existing) in the remaining 38 patients (Fig. 2).
When considering the pre-existing conditions (at least one of which worsened or appeared de novo in the group of the 76 patients), they occurred with statistically similar frequency in the 76 patients and In the 76 patients there were 137 complications (mean = 1.80 per patient), with 48 patients having experienced a total of 78 aggravations of at least one preexisting condition (mean = 1.77 per patient) and 52 having experienced the de novo appearance of a total of 59 abnormal conditions (mean = 1.13 per patient). These numbers are significantly worse in the subgroup of patients (n = 29) with at least one TSH measurements > 4.12 mU/L during exposure compared to the subgroup with none of such measurements above this threshold.
The conditions significantly linked to increased TSH levels appeared to be obesity, diabetes, hypertriglyceridemia and hypercholesterolemia ( Table 6). Because of changes in these first three conditions, there were 3 new cases of metabolic syndrome in the subgroup of the 29 patients (25/29 compared to 22/29 at baseline). In contrast, within the subgroup of the remaining 47 patients the cases of metabolic syndrome remained 10. Furthermore, in these 22 patients, a total of 47 criteria (mean = 2.14 per patient) required for the NCEP ATPIII diagnosis of metabolic syndrome did worsen, which compares with 10 criteria in the 10 patients (mean = 1.0 per patient) of the subgroup with Table 6 Details on the conditions that were considered "complications" if pre-existing (in the OFF-interfering drug period) and worsening (in the ON-interfering drug period) or not pre-existing (in the OFF-interfering drug period) and appearing de novo (in the ON-interfering drug period).* TSH ≤ 4.12 mU/L (data not shown in Table 6). Also, in the 38 patients who experienced no complications during exposure there were 10 with pre-existing metabolic syndrome. These 10 patients continued to have metabolic syndrome during exposure, but none of the NCEP ATPIII  Table 6.
(caption on next page) criteria worsened.

Discussion
There are strengths in our work, which counterbalance what at first sight is the limitation of being retrospective. Indeed, one advantage of the retrospective nature of this study was lack of any bias, such as increasing the frequency of monitoring clinically/biochemically patients while on the interfering drugs. Second, the 8 physicians share a careful methodology in recording themselves clinical, laboratory, instrumental, pharmacological data of their patients, whom they knew very well. This is different from extracting data from a database that had been entered by an unknown number of different persons with an unknown methodology. Third, for both the unexposed period and the exposed period each patient has at least two determinations of TSH, and in a given patient this minimum of 4 TSH assays were performed in the same laboratory. Fourth, we have evaluated a number of complications that are sensitive to changes in serum TSH, not just one or two.
There are some major findings of our work. First, approximately one in two hypothyroid patients taking tablet L-T4 therapy (391/730) also takes one or more medications that interfere with L-T4 intestinal absorption. When these patients under polypharmacy are selected to have a minimum of two TSH measurements prior to and during polypharmacy (exposure), then around one-third of them (41/114) have at least one TSH measurement above 4.12 mU/L during exposure, a rate that is only one-tenth at baseline. The second finding has to do with "complications", namely, the aggravation of pre-existing or the de novo appearance of any of the metabolic and cardiovascular conditions we have considered. Clearly, such complications are expected no matter what is the cause of elevated TSH. What we found, as a data previously unreported in the literature, is that those complication occur after a period of 24-55 months (median 31) of polytherapy in two-thirds of the tablet L-T4 replaced patients. During exposure, almost 40% of the patients with complications (29/76) had at least one TSH measurement > 4.12 mU/L compared with approximately 10% (7/76) at baseline. Third, the link between undertreated hypothyroidism (TSH > 4.12 mU/L) and complications is underscored by a sort of "carry-over" effect. Indeed, 24% (7/29) of the patients with complications who had at least one TSH measurement > 4.12 mUL during exposure also had at least one such measurement at baseline, which contrasts with the frequency of 8% (1/12) in the noncomplication group. Fourth, the said link is reinforced by the greater rate of complications in the subgroup of patients having at least one TSH measurement > 4.12 mU/L during exposure compared with the subgroup having all TSH measurements ≤4.12 mU/L during exposure (Table 5). Fifth, data are congruent by looking from the opposite perspective, namely absence of complications. Indeed, there were few patients (n = 5, with a total of 31 TSH measurements) who had no alteration in the biochemical or clinical parameters both during exposure and at baseline. Of the 31 TSH measurements, none was greater than 2.0 mU/L. When using the stringent threshold of 2.5 mU/L suggested by the National Academy of Clinical Biochemistry [6], the frequency of serum TSH levels above target continued to be significantly greater than in patients without complications (35% vs 17%, OR = 1.9), then one should conclude that a "desirable" level of serum TSH should be, indeed, below 2.5 mU/L. Several studies have found significant unfavourable relationships between increases in serum TSH, even within the reference range, and parameters that we have considered to indicate complications [9][10][11][12][13][14][15][16][17]. For instance, the large studies from Norway have found progressively greater levels of total cholesterol, LDL-cholesterol, triglycerides [14], SBP and DBP [15,16] along progressively greater bands of circulating TSH. In our study, it is noteworthy that the average TSH levels during exposure were 3.44 ± 4.08 mU/L (median 2.10) in the 76 patients who experienced complications but 1.58 ± 1.98 (median 1.28) in the 38 patients who did not. These two exposure average levels of TSH represent a great and a negligible increase compared to the corresponding baseline average levels (1.18 ± 1.44 mU/L [median 0.86] and 1.43 ± 1.09 [median 0.97]). In sum, the interference on the intestinal absorption of tablet L-T4 caused by certain medications, moves TSH from one class of metabolic/cardiovascular risk towards classes of greater risk.
Given the retrospective study design, association of elevated TSH and increased metabolic/cardiovascular risk could be due to a possible suboptimal compliance with medications in general that should be kept in mind. In addition, obesity and weight gain is associated with higher L-T4 requirements, thus a not fully uptitrated L-T4 dose could have led to increased TSH and metabolic/cardiovascular consequences [30].
Based on recommendation no. 13 of the 2012 ATA/AACE Guidelines, "patients being treated for established hypothyroidism should have serum TSH measurements done at 4-8 weeks after initiating treatment or after a change in dose. Once an adequate replacement dose has been determined, periodic TSH measurements should be done after 6 months and then at 12-month intervals, or more frequently if the clinical situation dictates otherwise" [1]. No recommendation on periodic TSH measurements are given by these [1] and subsequent guidelines [2] for hypothyroid patients who are under medications that increase thyroid hormone requirement. Based on the responses of almost 900 endocrinologists to a survey [29], after achieving stable target TSH levels, 55.5% of respondents recheck thyroid laboratory tests at 6-month intervals and 34% at 12-month intervals. Furthermore, once target TSH levels are achieved, approximately one-third of respondents (but half, if European) return the patient to the primary care physicians. In view of our data, hypothyroid patients under medications that affect thyroid hormone absorption/metabolism deserve more frequent rechecks.

Conclusion
L-T4 treated hypothyroid patients who start taking any medications known to cause undertreatment of hypothyroidism should mention this adjunctive therapy to the physician who is managing their hypothyroidism. Periodic measurements of TSH cannot be sporadic but relatively frequent, and it should be complemented by appropriate evaluation of pre-existing or de novo appearance of metabolic and cardiovascular disorders. Future guidelines should enforce this more comprehensive strategy for following up this category of hypothyroid patients.

Declarations of interest
None.

Appendix A. Supplementary data
Supplementary data to this article can be found online at https:// doi.org/10.1016/j.jcte.2019.100189. Fig. 3. Number of conditions per patient that were considered as complications, either worsened (if pre-existing) or appeared de novo, upon stratifying patients based on serum TSH levels (on target [≤4. 12 mUL] or not on target [ > 4.12 mU/L]). Graphically illustrated here are data reported in the last raw of Table.