The study included 148 patients with axSpA. Their mean age was 40 ± 11 years, mean BMI was 26.51 ± 5 m2/kg, mean disease duration was 8.3 ± 6.7 years and mean delay in treatment since diagnosis was 6.32 ± 6.37 years. Of these, 68.9% were males and HLA B27 was positive in 45.3 %. Disease subtypes were as follows: AS were 113, axial Psoriatic arthritis (PsA) were 27 and IBD with axial manifestations were 8, with the non-radiographic type constituting 39.1%. As for mean ASDAS-CRP initially before starting treatment, it was 3.65 ± 0.72.
As shown in figure 1 and table (1), the initial biologic taken was IL17-i in 103 patients, and TNF-i in the remaining 45. Non-switchers were only 55 patients (40 of whom were on IL17-i and 15 were on TNF-i), while switchers to a second line were 93 (62.84%). Among the switchers, 67 switched only once to a second drug line, 18 switched twice (12.16%) to a third line and only 8 switched further (5.41%) to a fourth line.
There was no difference between switchers and non-switchers as regards age, gender, smoking, BMI, disease duration, delay in treatment, HLA-B27, radiography of SIJ, baseline ASDAS and associated comorbidities. There was a statistically significant preponderance of patients with elevated faecal calprotectin among switchers.
Regarding rate of switching, those who switched once were 72%, twice were 19.4% and thrice 8.6% from total number of switchers. There was no significant difference between those who shifted once, twice or thrice as regards age, gender, smoking status, HLA-B27, BMI, delay in treatment initiation, radiographic classification of SIJ, associated comorbidities, initial ASDAS and first biologic used. As for disease duration (P=0.01), BMI (P=0.021) and elevated faecal calprotectin (P=0.000), they appeared to significantly increase the number of switching with a rate ratio of 1.05, 1.01 and 3.64 respectively. Those significant variables were then entered in multivariable poisson regression (table 2) to estimate the adjusted rate ratio (ARR). As a result, three factors were significantly associated with increased number of switches after adjustment for other factors: Elevated calprotecin (ARR= 3.65; P= 0.000), BMI (ARR = 1.025; P = 0.042), and disease duration (ARR = 1.02; P= 0.04).
Survival analysis:
Median drug survival time by treatment line (biologic order) for the initial, second, and third biologic lines (irrespective target cytokine) were compared and showed no significant difference (P= 0.136) between the three (15, 24 and 12 months respectively). (Figure 2)
Considering the type of therapy given during each line, the median survival of TNF-i was 12, 24, and 12 months when used as first, second or third line respectively, with one-year retention rate of 45.5, 60.3 and 37.8%. As for IL17-i, median survival was 18 months as first line, and 3 months as third line. When used a second line the median time couldn’t be calculated as only 2 cases switched at 3 months and the others were censored. IL17-i showed 60.4% one-year retention rate when used initially, 81.8% as a second line and 50% as a third line. No significant difference was found between the 2 therapies’ survival whether used initially as a first line or after failure of preceding lines (table 3, figure3; a, b, and c) (P>0.05).
For investigating the significant factors affecting time to switch to another treatment,(Andersen-Gill model (AG)) was conducted.
For the overall model,univariable regression revealed factors associated with higher risk of discontinuation and switching to the next line were higher BMI, as every unit increase in BMI was associated with 4.6% increased hazard of switching as well as elevated faecal calprotectin level which had a 3.15 higher hazard to switch compared to normal (HR = 3.15, p-value: 0.000). Patients with axSpA associated with PsA and IBD had a lower hazard to switch compared to pure axSpA (HR = 0.601 and 0.521 respectively).
Regarding IL17-i model and TNF-i model,the factors associated with higher risk of discontinuation and switching to the next line were higher BMI and elevated faecal calprotectin level (P<0.05). However, Patients with PsA were associated with lower hazard to switch compared to axSpA in IL17-i model(HR = 0.55) and Patients with IBD in TNF-i model(HR = 0.53).
Regarding Multivariable adjusted cox regression (AG model): (table 4)
For overall model:
Elevated faecal calprotectin level was significantly associated with lower survival, higher risk of discontinuation and switching to the next line, after adjustment for other variables. Patients with PsA and IBD had a lower hazard to switch compared to axSpA (HR = 0.58 and 0.33 respectively), after adjustment for other variables.
For IL17-i model:
Elevated faecal calprotectin level was significantly associated with lower survival of IL17-i, higher risk of discontinuation and switching to the next line, after adjustment for other variables (HR: 3.5; P: 0.000).
Patients with PsA had a lower hazard to switch compared to axSpA (HR = 0.53; p: 0.002), after adjustment for other variables.
For TNF-i model:
Elevated faecal calprotectin level and higher BMI were significantly associated with lower survival of TNF-i therapy, higher risk of discontinuation and switching to the next line, after adjustment for other variables (HR= 4.36 and 1.04 respectively). Every unit increase in BMI was associated with 4% increased hazard of switching.
Patients with IBD had a lower hazard to switch compared to axSpA (HR = 0.39; p: 0.004), after adjustment for other variables.