[guided]Why does this step matter? To iterate the Cauchy formula, we need the integral in $\zeta_1$ to produce a function that is holomorphic in the remaining variables. If the integral somehow destroyed holomorphicity, we could not proceed.
For each fixed $\zeta_1$ on the circle $|\zeta_1 - a_1| = r_1$, the point $(\zeta_1, z_2, \dots, z_n)$ lies in $\overline{\mathbb{D}^n(a,r)} \subset \Omega$ whenever $(z_2, \dots, z_n) \in \mathbb{D}^{n-1}(a', r')$. Since $f \in \mathcal{O}(\Omega)$, the map $(z_2, \dots, z_n) \mapsto f(\zeta_1, z_2, \dots, z_n)$ is holomorphic on $\mathbb{D}^{n-1}(a', r')$. The denominator $\zeta_1 - z_1$ is independent of $z_2, \dots, z_n$ and nonzero (since $|\zeta_1 - a_1| = r_1$ while $|z_1 - a_1| < r_1$), so
\begin{align*}
h_{\zeta_1}: \mathbb{D}^{n-1}(a', r') &\to \mathbb{C} \\
(z_2, \dots, z_n) &\mapsto \frac{f(\zeta_1, z_2, \dots, z_n)}{\zeta_1 - z_1}
\end{align*}
is holomorphic in $(z_2, \dots, z_n)$.
To pass holomorphicity through the integral, we use differentiation under the integral sign. Fix any compact polydisc $\overline{\mathbb{D}^{n-1}(a', s')} \subset \mathbb{D}^{n-1}(a', r')$ with $s_j < r_j$ for each $j \geq 2$. On the product $\{|\zeta_1 - a_1| = r_1\} \times \overline{\mathbb{D}^{n-1}(a', s')}$, the function $f$ is continuous (as a restriction of a [holomorphic function](/page/Holomorphic%20Function) to a compact subset of its domain), hence bounded, say by some $M > 0$. The denominator satisfies $|\zeta_1 - z_1| \geq r_1 - |z_1 - a_1| > 0$ uniformly for $z$ in the compact polydisc. Therefore the integrand is uniformly bounded, and differentiation under the integral sign with respect to the Wirtinger operator $\partial_{\bar{z}_j}$ for $j \geq 2$ yields
\begin{align*}
\partial_{\bar{z}_j} \left[\frac{1}{2\pi i} \oint_{|\zeta_1 - a_1| = r_1} \frac{f(\zeta_1, z_2, \dots, z_n)}{\zeta_1 - z_1}\, d\zeta_1\right] = \frac{1}{2\pi i} \oint_{|\zeta_1 - a_1| = r_1} \frac{\partial_{\bar{z}_j} f(\zeta_1, z_2, \dots, z_n)}{\zeta_1 - z_1}\, d\zeta_1 = 0,
\end{align*}
since $f \in \mathcal{O}(\Omega)$ implies $\partial_{\bar{z}_j} f = 0$. This confirms that the integral is holomorphic in $(z_2, \dots, z_n)$.[/guided]