**4. Discussion**

In the presented study, it was stated that the baseline 25(OH)D level had a profound impact on the changes observed after the intervention, so it may be concluded that the e fficacy of the applied dietary intervention of daily salmon intake depends on the baseline 25(OH)D level. In the total studied group, the median change after eight weeks of intervention was −8.1 nmol/L. Taking into consideration only the participants with adequate 25(OH)D level (≥ 50 nmol/L), it was also a drop (median = −6.7 nmol/L), whereas in participants with inadequate 25(OH)D level (< 50 nmol/L) after eight weeks of intervention, a median rise of 8.2 nmol/L was observed.

While comparing the obtained results with studies by other authors, it may be supposed that it is a general association—in groups with lower baseline 25(OH)D levels, a higher e ffect of intervention may be observed, so it may confirm that the baseline 25(OH)D blood level plays an important role. In a similar intervention study, lasting eight weeks, conducted in wintertime in Iceland, Spain, and Ireland, in which its participants (mean baseline 25(OH)D blood level 61.9 nmol/L) consumed around 450 g of salmon per week (compared to 350 g per week in our study), the mean rise of 25(OH)D was 8.4 nmol/L [40]. In another study conducted in Finland, on participants with a much higher mean baseline 25(OH)D level of 124.0 nmol/L, no significant increase in 25(OH)D was noted when consuming 300–600 g of fatty fish (including salmon) per week after eight weeks of intervention [41]. However, most studies of this type are small intervention studies with a limited number of participants, therefore comparing the obtained results with the conclusions from a meta-analysis summarizing the most essential results may be helpful. According to a meta-analysis of Lehmann et al. [29], the mean change in 25(OH)D for interventions lasting 4–8 weeks was 3.8 nmol/L; whereas in intervention groups with an inadequate mean 25(OH)D baseline (< 50 nmol/L), the change was 6.1 nmol/L; and in groups with adequate (≥ 50 nmol/L) 25(OH)D blood levels at baseline, the change was only 3.9 nmol/L. Taking into account the results of various studies, presented above, it may be confirmed that the most important factor for the e fficacy may be the baseline 25(OH)D level.

However, in the presented study in the total study group, a decrease of 25(OH)D level after four weeks of intervention (phase I) was stated and later on (phase II), an increase (compared to the level after four weeks) was observed. The decline is very surprising, as salmon is known to be a good source of vitamin D [22,26], and in numerous studies, an increase in 25(OH)D was shown, also only after a four-week-long intervention of salmon intake in the winter [29]. Nevertheless, a recent intervention study conducted in Norway indicated that even a weekly consumption of 750 g (compared to 350 g in the presented study) of salmon for eight weeks was not su fficient to prevent the 25(OH)D decrease [42]. In that study, 25(OH)D decreased both in the salmon intervention and the control group, however, the decline was significantly lower in the salmon group compared to the control group.

What should be underlined is that the intervention period in the referred study [42] was from August/September to October/November, whereas in the presented study it was later—from October to December. Therefore, the decline seen in the presented study from week 0 to week 4 may correspond to the one observed in the Norwegian study (after eight weeks of intervention). In the presented study, the amount of salmon consumed was more than two times less than in the Norwegian one, and still in the second part of the intervention (from week four to week eight) an increase in 25(OH)D was observed. Unfortunately, to our best knowledge, the reason why such high amounts of dietary vitamin D from fish do not contribute to an immediate increase of 25(OH)D is not ye<sup>t</sup> described or well known. There are only some hypotheses that could be listed.

The decrease may be attributed to a lack of adequate sunshine exposure to cause skin vitamin D synthesis during the time of intervention. From other studies, it is known that in countries such as Poland, skin synthesis of vitamin D is observed only from April to October (if the sunshine exposure is adequate) [20]. According to a recent Polish study [43], the highest levels of 25(OH)D in Poland are observed in August and the lowest in January, which could be explained by insolation and 25(OH)D synthesis during summer and the mobilization of vitamin D stored in the body (during the summer months) in the winter.

This hypothesis corresponds to the initial decrease of 25(OH)D levels in the presented study, which was performed from October to December. Therefore, it may be assumed that in phase I (until week four), not dietary vitamin D, but rather previously stored vitamin D had been used by the organism causing the constant decrease of 25(OH)D blood level, but in phase II (from week five to week eight), in which not enough vitamin D was stored in the body anymore, the daily intake of 50 g of fish and dietary vitamin D provided must have been intensively stored and metabolized to improve its status. Other studies also emphasize that it is the vitamin D storage from sun exposure and its release from fat tissues in winter that is the major factor contributing to 25(OH)D levels throughout the year [44], and dietary vitamin D is associated with 25(OH)D mainly in the winter and spring [45], in which perhaps not that much vitamin D is left in fat tissue and, therefore, the body must depend on vitamin D from food sources. This could also be the reason for the observed impact of baseline vitamin D status (adequate or inadequate) on 25(OH)D changes throughout the study, since in participants with inadequate vitamin D status, a lower decrease in phase I and a higher increase in phase II were seen, compared to participants with adequate vitamin D status at baseline.

Another possible explanation could be the di fference in dietary vitamin D intake throughout the study. The statistical comparison of participants with inadequate (<50 nmol/L) and adequate (≥50 nmol/L) 25(OH)D levels at baseline showed that at week four (t4) participants with adequate (≥50 nmol/L) 25(OH)D levels had a close to significant (*p* = 0.0515) higher vitamin D intake (median = 10.65 μg) in the first part of the intervention (t0 to t4) compared to the ones with inadequate (<50 nmol/L) 25(OH)D levels (median intake = 9.73 μg). What should be noted is that the dietary vitamin D intake of participants with adequate (≥50 nmol/L) 25(OH)D levels covered the recommended 10 μg [11,25,26], which was also assumed in the study. Therefore, the possible reason for having an adequate 25(OH)D level at week four could be the higher (covering the recommended 10 μg) intake of vitamin D in the preceding time (from t0 to t4). At week eight, the vitamin D intakes did not di ffer significantly between groups (*p* = 0.3411), therefore the influence of vitamin D intake on di fferent outcomes at that time remains unclear.

There are also other possible explanations, such as levels of other nutrients, that are related to the metabolism of vitamin D [46]. It may have been influenced by a decreased calcium level prior to the study [47]. This level was not assessed in the presented study, but if this level is lowered, it may influence not only the 25(OH)D level, but also the 1,25(OH)2D level, and the activity of 25(OH)D 1- α-hydroxylase [47].

Therefore, it may be hypothesized that apart from baseline 25(OH)D levels, it is the starting time of intervention and the dietary intake of vitamin D and other nutrients that play a big role in the outcomes of such short-term fish intake interventions. Other hypothetical possible explanations to such intervention outcomes could be that the digestive system needs to adjust to utilize vitamin D from the smoked salmon. Another could be that the metabolism and storage of vitamin D are faster in the case of high dietary supply (intervention) while the levels of vitamin D stored from the summer exposure are also high, which might lead to lower 25(OH)D levels because the whole consumed amount of vitamin D is converted to other metabolites in a dynamic way. Last but not least, the vitamin D metabolism might have other various pathways depending on the baseline level, exposure, dietary intake, and applied intervention, which we are not ye<sup>t</sup> aware of.

Based on the meta-analysis of Lehmann et al. [29], it may be indicated that the length of the intervention also plays a key role—short-term studies (4-8 weeks) revealed a mean di fference of 3.8 nmol/L and long-term studies (around six months) revealed 8.3 nmol/L [29]. If the interventions last longer, there is more time for a change of 25(OH)D. Therefore, it may be supposed that prolonging the conducted intervention would have resulted in a higher increase of 25(OH)D in the studied group. It may be hypothesized especially for participants with an adequate 25(OH)D level for whom after eight weeks of intervention a lower 25(OH)D serum level (median 56.8 nmol/L) than at baseline (median 70.4 nmol/L) was still observed, but higher than after four weeks of intervention (median 45.4 nmol/L), so a progressive increase may be supposed.

In spite of a number of arguments for choosing salmon as a vitamin D source in diet, namely its availability, as well as the content of nutrients and other components, the price of this product is quite high [48], contributing to low consumption in Poland (in 2017: 0.63 kg/person/year) [48]. There are other much cheaper (and therefore consumed more often) fish species available in Poland with significant vitamin D and combined EPA and DHA content, as well as low mercury and dioxin levels, that should be recommended, such as herring and rainbow trout. Both those species have very high (>15 mg/g) combined EPA and DHA content, low (≤ 0.1μg/g) mercury levels, as well as quite low (0.5–4 pg TEQ/g) dioxin content, so according to the FAO and WHO [32] similarly as in the case of farmed Atlantic salmon, consuming them has more benefits than risks. Their prices in Poland are lower than salmon and they are as follows: rainbow trout—23.62 PLN/kg (approx. 5.42 €/kg) and herring—16.03 PLN/kg (approx. 3.68 €/kg), compared to 57.90 PLN/kg for salmon (approx. 13.28 €/kg) [48].

Moreover, herring and rainbow trout contain significant amounts of vitamin D: 19.00 μg and 13.50 μg/100 g, respectively [22]. There were some single intervention studies in which participants consumed 750 g/week of herring for six weeks [49] or 400–600 g/week of fatty fish including herring and rainbow trout for eight weeks [42] which did not reveal any significant effect of the applied dietary intervention on total 25(OH)D. What should be pointed out is that the study groups consisted of only 32 [49] and 11 participants [41]. Therefore, similar intervention studies with herring or rainbow trout should be conducted in larger study groups in order to explore the influence of consumption of those species on 25(OH)D blood levels. If they also have a positive influence on vitamin D status, it could have a greater impact on Polish people to recommend them to eat more of those fish species, as they can afford and already consume them in higher amounts than salmon (herring 2.56, trout 0.50 kg/person/year in 2017) [48].

According to the American Heart Association [50], as well as the Polish National Food and Nutrition Institute [51], the recommended fish (especially fatty fish) intake is at least two times a week. Fatty fish include the fish species mentioned above such as salmon, herring, and rainbow trout, which are good sources of vitamin D, and should not be reduced from the diet, as they contain less mercury and dioxins than other species. From other studies, it is known that fish intake is the most influential food source contributor to vitamin D intake [52,53], and there may be a strong correlation between fatty fish intake and 25(OH)D levels [54]. A recent cross-sectional Norwegian study revealed that for a median vitamin D intake of 10.3 μg, the mean 25(OH)D level was 64.0 nmol/L, while the prevalence of deficiency (defined as 25(OH)D < 50 nmol/L) was only 24.7%; in such a situation, a low vitamin D deficiency prevalence was observed, even in winter [55]. This suggests that with an adequate vitamin D intake (at least 10 μg/day) it is possible to achieve an adequate 25(OH)D level, maybe even in seasons with limited sunshine exposure such as the autumn and winter seasons in Poland. It corresponds with the obtained results, as even as short a period as eight weeks of dietary intervention of adequate vitamin D intake influenced the 25(OH)D levels and was revealed to be a promising option.

The novelty of the study is the fact that it was the first fish intake intervention study to asses 25(OH)D levels in a Polish population. Moreover, it is one of the first fish intake intervention studies in which a decrease in 25(OH)D was observed despite high salmon intake. The hypothetical explanations for that are listed above. Nevertheless, this indicates that this matter needs further study.

Although the presented study was the first fish intervention study conducted in Poland to access the efficacy of such intervention on 25(OH)D serum levels, limitations of the study should be indicated. The most important issue is associated with the fact that in the presented study there was no control group with no dietary intervention, which makes it more challenging to draw conclusions. However, this study is not the first one to analyze the influence of fish intake intervention on vitamin D status (measured as total 25(OH)D). Similar studies were conducted in other countries (but not in Poland), such as one in Iceland, Spain, and Ireland [39], with similar dietary intervention and similar results. Thus, it can be hypothesized that in the present study, the reason for the recovery of 25(OH)D was also due to fish intake and the presented study is of grea<sup>t</sup> importance, despite the lack of a control group.

Taking it into account, it must be stated that this matter needs further study, as there are not many studies conducted on that topic, and, what has to be underlined, there were no such studies in Poland, so far. Finally, the studies conducted in other countries are frequently carried out on very specific participants such as prisoners [56], sex o ffenders [57], overweight individuals following a low-calorie diet [39], and 8-9-year-old children [58], therefore, there is a grea<sup>t</sup> need for similar intervention studies (conducted using various, but defined fish species) on larger, homogenous study groups covering various ages for both male and female participants.
